This PDF file contains the front matter associated with SPIE Proceedings Volume 12550, including the Title Page, Copyright Information and Conference Committee lists.

With the development of robot technology, automatic assembly technology based on industrial robots has been greatly applied. Considering the high-precision automatic grasping of heterogeneous products, robot grasping detection technology is put forward. Although the current method of using camera is relatively mature, it is greatly affected by the ambient light and the surface material of the object, resulting in the limited versatility. Since the 3D structured light method based on the laser scanning galvanometer has the characteristics of high speed and high accuracy, strong environmental adaptability and weak influence by the surface material of the object, the structured light equipment using laser scanning galvanometer was installed at the end of the robot to realize the grasping detection of the products. Through hand eye calibration, point cloud acquisition, point cloud processing, point cloud matching and other links, the point cloud collection and accurate matching of local features are realized. The experiment shows that the grasping detection accuracy is less than 0.32mm. It can not only achieve accurate grasping of different products, but also has the characteristics of strong adaptability to surface materials, environmental lighting, and high accuracy.

The multi-frequency heterodyne is a temporal phase unwrapping algorithm that is commonly used in phase-shifting methods. The main principle is that two phase maps with similar periods are differenced to obtain a reference map with a longer period and no ambiguity. However, this method poses a noise problem: the heterodyne procedure leads to the accumulation and amplification of errors in the absolute phase maps, which may result in fringe order jumps. In this paper, a novel multi-phase fusion strategy is proposed, which incorporates correlation constraints in the phase decoding process to reduce the effect of noise. First, a priori constraint on the background illumination and modulation intensity of the multi-frequency fringe pattern is proposed, and a phase optimization function is established by correlating the phase signal models of different frequencies. Since the optimal value of the optimization function is localized, the optimization process is divided into two parts: phase order correction and local phase optimization; then, anomalous phases are identified and denoised according to phase anomalies and the priori condition failures. Finally, the effectiveness of the proposed strategy is verified quantitatively on a standard plane: the SNR of the optimized 4-step phase-shift method is close to that of the 16-step phase-shift method. Furthermore, the strategy also demonstrated a significant reduction in the phase jump problem and an improvement in the SNR of the phase map in 3D PCB measurement experiments.

n industrial application of structured light systems (SLS), besides the measurement accuracy, the robustness in long-time working situations is also significant. This requirement is hard to be satisfied due to the drifting of system parameters caused by factors such as the generated heat of the devices and the temperature variation of the environment. To ameliorate this drawback, this paper proposes a re-calibration method of SLS parameters in temperature-varying situations based on utilizing a series of circle marks as control-points. In this method, after initial calibration, a measurement of the whole workspace including these control-points is carried out as soon as possible, in which a series of two-directions phase images and an extra white image are captured. Through phase calculation and reconstruction, the precise poses of the control-points can be obtained and saved. Then, cost function is constructed by utilizing these control-points based on bundle adjustment and the system parameters are optimized. Compared with previous re-calibration methods, since extra constraint of relative pose between the camera and the projector is provided by registered circle marks, a target object with ground truth used to calculate the scale information in previous methods is not necessary anymore, which improves the compatibility of SLS application. Experiments are also carried out to verify the performance of this method.

Point cloud registration is the key step of 3D non-contact precision measurement for complex mechanical parts based on machine vision. Due to extreme dependence on initial position, slow iterative convergence and many wrong corresponding point pairs, Iterative Closest Point (ICP) algorithm could not satisfy the requirements of point cloud registration efficiency and precision for measurement of large quantities of complex mechanical parts, so an improved ICP method of measurement point cloud registration for complex mechanical parts by combining ICP with Intrinsic Shape Signature-Fast Point Feature Histogram (ISS-FPFH) feature is proposed. In order to reduce the registration quantity of the point cloud and keep the original subtle features on surface of point cloud, the voxel filter based on the point close to center of gravity is proposed to preprocess for point cloud down sampling. It is difficult for traditional ICP algorithm to determine the appropriate initial position and will lead to the registration failure of multi-view measurement point cloud, the Sample Consensus Initial Registration (SAC-IA) algorithm based on ISS-FPFH is applied for coarse registration. In order to solve the problems of slow iterative convergence and many wrong corresponding point pairs of traditional ICP algorithm, the point-to-plane ICP algorithm by combining with normal vector angle constraint is proposed for fine registration. Through experimental comparison and analysis, the results show that the accuracy and efficiency of the method proposed in this paper can meet requirements of 3D non-contact precision measurement for large quantities of complex mechanical parts.

Calibration is essential for the three-dimensional light field endoscope, and the two-step calibration method based on line features was proposed by us to accomplish it before. In the second step of the two-step calibration method, the relationship between the projections of virtual feature points on the microlens image and the central sub-aperture image was used to calibrate the parameters about the microlens array, but we find that the feasibility of the calibration method has a depth constraint on the border of the checkerboard. In this paper, we deduce and demonstrate this constraint, and an optimization algorithm is designed to improve the accuracy of the second step calibration due to the inaccurate black-and-white boundary detection. Experimental results show that our method is effective and accurate for the calibration of the three-dimensional light field endoscope.

Panoramic 3D measurement becomes increasingly important for fringe projection profilometry (FPP). Traditional physical markers-assisted (PMA) method suffers from inefficiencies and non-complete measurement. An optical markers-assisted (OMA) panoramic 3D method has been recently proposed, which enables accurate, efficient and non-destructive panoramic 3D measurement. In this paper, we give a comprehensive comparison between OMA and PMA, which provides reasonable suggestions for different panoramic 3D measurement applications.

ynamic 3D face acquisition has been widely used in the fields of real-time expression capture, face recognition, plastic surgery and film entertainment. High-precision 3D shape can be obtained by fringe projection 3D measurement technology. However, the phase ambiguity is one of the key issues must be solved. According to the features of human face, a method based on feature point matching and active binocular vision is proposed to address this problem. The specific method is as follows: 1) Project three sinusoidal fringes with a phase difference of 2π/3. Capture the deformed fringes by the left and right cameras simultaneously. 2) According to the phase shift method, calculate the corresponding intensity images and the wrapping phase maps from the deformed fringes captured by the left and right cameras respectively. 3) Detect the face feature points in the left and right intensity images respectively. 4) Create the rough face disparity map by matching the feature points of human face in the left and right intensity images. 5) On the basis of rough disparity map, obtain the dense and high-precision disparity map by directly matching the wrapping phase of the left and right images. And then reconstruct the 3D face. Compared with the traditional method, the proposed method does not need the spatial phase unwrapping, nor does it need the projection of additional patterns to assist the phase unwrapping, which shortens the acquisition time while ensuring the measurement reliability. A DLP projector and two cameras with resolution of 1.3 megapixels and speed of 210fps were used to build an experimental system to verify the proposed method. The experimental results show that the proposed method is robust and the 3D face acquisition speed reaches 70fps.

High-accuracy 3-D shape measurement of turbine blade is important for manufacturing engine and aircraft’s safety. At present, the most commonly used three-coordinate measuring machine in the industrial field may cause scratches on the surface of turbine blade. Optical 3-D measurement methods, such as fringe projection profilometry, are unable to carry out targeted measurement of geometric structures with different levels of precision. This paper presents an automatic measurement method based on pneumatic gripper and 6-axis robot. The proposed composite system includes global sensor with large scanning range and local sensor with high accuracy, which can perform 3-D reconstruction with specific resolution for different positions of turbine blade. We use a 6-axis robot and pneumatic gripper to grasp the turbine blade automatically. The 3-D scanning is completed according to the viewpoints planned in advance. The point cloud data of different angles are aligned to obtain a high precision 3-D model of turbine blade surface. The experimental results show that the proposed automatic scanning method can effectively avoid the measurement error caused by manual grasping operation. On the basis of complete 3-D measurement of the whole surface, more dense point cloud data are obtained for fine structures such as blade edge. Our result is of great significance to the field of industrial measurement.

In the field of 3D measurement, fringe projection profilometry attracts the most interest due to its high precision and convenience. However, it is still challenging to retrieve the unambiguous absolute phase from a single fringe image. In this paper, we propose a deep learning-based method for retrieving the absolute phase of triangular-wave embedded fringe images. Through the learning of a large amount of data, we use two neural networks to obtain high-precision wrapped phase and coarse absolute phase from the triangular-wave embedded fringe images respectively so as to obtain accurate fringe order. Combining the wrapped phase and fringe order, we can obtain high-precision absolute phases. The experimental results demonstrate that compared with our previous proposed composite dual-frequency fringe coding strategy, the fringe image of the new triangular-wave embedded fringe coding strategy as the input of the network can obtain the absolute phase with higher accuracy.

In this paper, ultrafast laser ablation of Aramid Fiber Reinforced Plastic (AFRP) and its application in satellite reflector antenna was studied. Intrinsic properties for transmission spectrum were measured. For material processing the influence of pulse energy fluence, pulse overlapping rate and laser wavelength (from the NIR to the NUV) on appearance qualities (namely the blackening of cutting edge) were studied. It was shown that the wavelength played an important role in controlling the machining defects such as blackening. While NUV wavelength long pulse (10 ps) laser generated a smooth cutting cross-section with material’s intrinsic colors, the NIR wavelength resulted in a carbonization-prone ablation, even if the femtosecond laser was considered. The possible mechanisms involved were discussed. The application of laser trimming of satellite antenna was demonstrated.

Coded Aperture Snapshot Spectral Imaging (CASSI) is an effective tool to capture spectral images, which has the advantages of snapshot imaging, high luminous flux, high signal-to-noise ratio and low sampling frequency. However, conventional CASSI generally uses refractive prisms or gratings for spectral dispersion, which leads to the nonlinear dispersion phenomenon and the requirement of large detector chip respectively. To overcome these issues, conventional refractive prisms or gratings are replaced by an axially dispersive diffractive optical element (DOE, i.e., computational optics) together with a RGB Bayer filter (i.e., a color-coded aperture) in this study. Specifically, the spatial-spectral information of a test scene is jointly modulated by the DOE and the Bayer filter integrated with a sensor chip. A fully differentiable imaging model is built based on the principle of diffractive optics and the deep learning technology. Furthermore, an optimization design of the DOE with the coded aperture is realized through an end-to-end approach, the output spectral images of which are restored by a Res-Unet neural network. Several simulation results show that up to 31 high-fidelity spectral bands in the range of 400 to 700 nm with a good spatial and spectral resolution can be recovered by the proposed snapshot system.

Polarization cameras integrated with conventional four-channel Micro-Polarization Arrays (MPA) are usually used as phase shift devices in focal plane-type interferometric systems to achieve simultaneous phase shifts. However, such phase shift devices suffer from low sampling rates and a lack of mature resolution reconstruction techniques to resolve pixel-level alignment errors. To solve such problems and further improve the phase reconstruction accuracy of the focal plane interferometric system, this paper proposes to apply the "Bayer-like" three-channel MPA to the interferometric system to realize the simultaneous phase shift. The arrangement of the three-channel MPA can reduce the processing complexity of the element and improve the inter-channel sampling rate. In addition, this structure can better utilize the resolution reconstruction idea of Color Filter Array (CFA) to solve the inherent pixel-level alignment error problem between simultaneous phase shift interferograms. Taking lateral shearing interferometry as an example, the process of acquiring simultaneous phase-shifted interferograms and phase reconstruction in an interferometric system with three-channel MPA is verified by simulation. The experiment results shows that the proposed three-channel MPA structure performs better than the traditional four-channel MPA structure in terms of visual reconstruction effect and quantitative index of phase reconstruction.

Accurate data transmission is important in biological applications and biomedical devices. Realizing high-fidelity optical wireless transmission with a large penetration depth is desirable. Here, a sequence of amplitude-only patterns at the transmitter is considered as information carrier to encode the transmitted data. The total light intensity at the receiving end is collected and recorded by a bucket detector. We verify feasibility and effectiveness of the proposed approach with different thicknesses of a biological sample. The experimental results demonstrate our proposed method based on zero-frequency modulation using low light intensity.

Shallow underground target detection technology has received more and more attention, such as pipeline detection, cavity detection, underground military facility detection and underground archaeology. Based on the description of the acoustic-to-seismic coupling mechanism, the technology of laser speckle interferometry and its characteristics for acoustic-to-seismic coupling detection methods are summarized and analyzed. The subwoofer based on the sound catheter is used as the sound source excitation module, and the laser speckle interferometry system is used to conduct rapid detection experiments on the vibration characteristics of the ground surface coupled by sound waves, and the changes of laser speckle interference fringes are studied under different experimental conditions such as sound wave intensity and buried environment. Experimental results show that the laser speckle interferometry can be applied to the rapid measurement of acoustic-to-seismic coupling characteristic signals and has broad application prospects in the field of rapid non-destructive detection of flexible shallow buried objects.

Shape-From-Focus (SFF) is a technique that can obtain 3D features of objects through monocular cameras, it can be applied with high precision in optical microscopy. However, since SFF relies heavily on the texture information of the object's surface, when the texture of some regions in the object is poor, it always leads to low accuracy or even serious errors in the depth map reconstruction of this region. Most research in SFF has focused on finding and eliminating these low-texture regions, but for the eliminated regions, how to accurately recover the depth information is not well addressed. Less attention has been paid to enriching the texture information of these regions. Considering that bright-field and dark-field light sources often present images with different textures in microscopy, this paper proposes a method to enrich image textures by fusing bright-field and dark-field images. The proposed method needs to acquire two sets of image sequences in bright field and dark field respectively, and fuse the images based on wavelet transform at the same index position in the two sets of image sequences, so that the fused images sequence can retain the high frequency components in both bright and dark field images, the fused set of image sequences is then used for SFF to calculate the depth map. Experimental results with both simulated and real objects are presented to validate the proposed schemes.

Surface temperature of explosive fireball is vital for the evaluation of the thermal damage ability. In order to improve the spatial and temporal resolution of the measurement, based on the thermal radiation law, the principle and limitation of the equipment used to measure the surface temperature of explosive fireball are analyzed, such as infrared thermography camera, photoelectric pyrometer etc. Based on CMOS image sensor and its built-in filter array, a temperature measurement method based on visible band is proposed to realize the imaging measurement of the surface temperature of explosive fireball. An experimental verification was carried out by using infrared thermography camera and a high-speed camera-based pyrometer, the experimental results show that there are differences in the shape of fireball measured by different temperature measuring systems, the reason of the result is analyzed. The research has laid a foundation for the construction of infrared and visible band combined temperature measuring system.

The heat flux of the explosion field is one of the important parameters to characterize explosion thermal damage power, and the dynamic response characteristics of the heat flux sensor is crucial to accurately measuring the heat flux in explosion field. In order to improve the dynamic characteristics of the heat flux sensor used in explosion scenes, the simulation optimization of the heat flux sensor based on infrared detection is carried out. Based on the establishment of the theoretical model of the heat flux sensor, the thermal response of the sensor sensitive element under heat flux load is analyzed by numerical simulation, and the spectral analysis of the radiation from the back wall surface is carried out. The simulation results show that the response time of the sensitive element is reduced to 0.8ms after optimization, and the relative error between the heat flux calculated by the temperature of the back wall surface and the incident heat flux is only 1.75%. The spectral analysis results of the back wall surface provide a basis for the selection of the coating and infrared temperature measurement devices. The heat flux sensor is tested in the calibration experiment, and the experimental results show the rationality of the optimization of the sensitive element

In fringe projection profilometry, by taking the projector as a camera, depth measurement has the similar principle with binocular vision. The matching of feature points carrying depth information plays an important role in measurement. Due to the existence of difficultly-corrected projector errors, the measurement results of using a fixed point in camera corresponding to the points in the projector or of conversely using a fixed point in projector corresponding to the points in the camera is not exactly same. According to fringe projection system, a function representing the relationship between the depths and the corresponding fringe phases is introduced for measurement. This traditional method usually requires three or more known reference planes to calibrate parameters determined by the measurement system. However, another way is also feasible. Based on the measurement system, a generic function representing the relationship between the depths and the corresponding pixel shifts are also presented in this paper. Using these two functions allows us to recover the depth information of the object. Theoretical analyses are carried out to investigate the performances of these two methods. Comparative experiments are implemented, and their results demonstrate that the depth measurement method from pixel shifts is immune to the influence of high order harmonics as a little expense of efficiency, being better than the method from fringe phases having high measurement efficiency regardless of the low accuracy.

The ice cream stick is a kind of wood product with a plane surface. For the qualified stick it has the nearly flat surface. However, due to slender and thin shape the bend defect is prone to occur during machining of ice cream stick. At present, manual technique is often used to identify the bend defect in the industrial fields, which holds many problems such as low efficiency, low automation and unreliable detection result. In the paper, the PCA-based bend feature selection method is proposed which is applied in defect detection of ice cream stick. At first, the 3D data of object surface is obtained based on the three-dimensional (3D) measurement principle of line structured light. Then four kinds of bend features are designed according to the shape characteristics of the light stripe, which include variation coefficient, correlation coefficient, determination coefficient and straightness metric. At last, the bend feature selection is operated based on the PCA-based method. The research results provide valuable reference for the engineering application in the intelligent defect detection.

Limited by the imaging dynamic range of the camera, the phenomenon of over-exposure and over-dark often occurs in the 3D measurement of strong reflective sheet metal parts, resulting in incomplete measurement result. One of existing methods such as multiple exposure can measure most of the visible area under a single viewpoint, but the visible area with too small or too large incidence angle still cannot be measured. To solve this problem, in this paper, a method of viewpoint planning for sheet metal parts with strong reflection is proposed. The method introduces the surface reflection model of reflective sheet metal parts into viewpoint planning to achieve the synchronous optimum of measurement efficiency and data integrity. Firstly, according to the measurable region of the surface structured light 3D measurement system and CAD model, the candidate viewpoint set is randomly generated in the sampling space, and the visibility matrix is constructed by analyzing whether each candidate viewpoint is visible to each patch of the model. Then, the surface reflection model of sheet metal parts with strong reflection is constructed, and the reflection coefficient of the visible patches under each viewpoint is calculated according to the reflection model. Based on this, the measurability of the visible patches of the viewpoint under multiple exposures is calculated, and the visibility matrix is updated. Lastly, through the viewpoint quality evaluation function constructed based on the data coverage increment and multiple-exposure time, the viewpoint with the highest quality is selected heuristically until the coverage requirement is met. Experiments show that the algorithm can improve the measurement efficiency and ensure the integrity of the measurement data.

Holes are common and important feature information in sheet metal parts and play a decisive role in product production and assembly. The 3D contour reconstruction method based on passive binocular vision has been widely used in hole measurement due to its high accuracy and robustness. It can obtain dimensional data such as diameter and center of round holes, but there are still problems in measurement accuracy and integrity, and it is difficult to accurately measure special-shaped holes. Aiming at these issues, a fringe projection-based 3D contour measurement method for holes on sheet metal parts is proposed. In this method, two constant movements are performed on the extracted contour curve to the surrounding area, and the 3D contour points can be reconstructed according to the phase matching. Therefore, the real 3D contour is reconstructed inversely according to the corresponding points of the previous two reconstructed contours. With this method, contours only need to be extracted on one image, so it can reduce the effect of inconsistency in contour extraction. In addition, the accuracy and integrity of the 3D contour can be guaranteed by using high-quality phase matching of the sheet metal area. The experimental results show that this method can achieve accurate and complete 3D contour reconstruction for sheet metal holes of various shapes, and the accuracy can reach 0.08mm.

To reduce the absolute position error, this paper mainly uses Regularized Parameter Identification (RPI) method to in-situ calibrate the kinematic model of a manipulator. Firstly, the model is built according to the MDH modeling method to get the parameter error model. Secondly, as a parameter identification process, the regularization algorithm is used to compensate all the errors on the model of the manipulator. Thirdly, a calibration experiment is carried out to verify the RPI method by using a laser tracker. The experiment results show that the uncalibrated absolute position error of the manipulator is 0.553mm for RPI and 2.533mm for that without calibration. It shows that the regularization algorithm can effectively reduce the absolute position error of the robot after calibration.

Transparent parallel flat optical components are found everywhere and used in a wide range of applications. Surface topography and inhomogeneity distribution are important physical parameters for the components. In order to improve the performance and reliability of components, topography should be measured with high accuracy. The weighted multi-step sampling algorithm based on wavelength-shifting interferometry is an effective method for multi-surface topography recovery, which is often based on the three-surface measurement and has poor accuracy for rear-surface measurement. The four-surface measurement can compensate for this drawback, but the system and the solving process are more complex. There are fewer studies on the analysis of algorithm effectiveness. In this paper, through theoretical derivation, it is proved that the validity of the four-surface measurement algorithm is only related to the recovery of three sets of interference signals. The distribution of the wavefront residuals RMS values under different combinations of phase shift coefficients and cavity length coefficients are used to summarize the algorithm error distribution law and the best matching database of phase shift parameters under four-surface measurement is established. The effective cavity length coefficients and phase shift coefficients are selected to compare the wavefront recovery under three-surface and four-surface, and it is verified that the measurement accuracy of the four-surface is better when effective parameters are selected.

The automatic 3D measurement technology based on surface structured light has been widely used in the production of metal sheet parts. Before the measurement, it is necessary to manually plan the robot's measurement viewpoint and path, and each viewpoint needs to take into account the measurement quality of the shape surface and holes at the same time. The planning is very difficult, and manual planning results in long offline debugging time and low measurement efficiency. To address this problem, this paper proposes a view planning method that focuses on holes of metal sheet part. Firstly, the relationship between view angle and hole reconstruction quality is constructed to analyze the influence of view angle on hole measurement. Then, candidate viewpoints, which meet the visibility and hole reconstruction quality requirement, are generated. Finally, the Markov decision process is used to model the coverage planning problem, and the Monte Carlo tree search method is used to generate the shortest viewpoint sequence which meets the coverage ratio requirements. The experimental results show that the method proposed in this paper can realize the complete measurement of complex surface features such as holes, and plan an efficient robotic measurement path.

This paper presents a phase-stitching method to increase the range resolution of a Frequency-Modulated Continuous Wave (FMCW) lidar. Based on an external-modulated Electro-Optic Modulator (EOM), the modulation range is increased when ten sidebands in the frequency domain are generated. By stitching the phase of these ten sidebands of the signal, the effective tuning range is broadened 55 times so that the ranging resolution is increased. To improve the detection precision, the system includes a reference coherent detection arm to correct the nonlinearity of the modulation source. The phase stitching method applied in this paper can achieve high-ranging resolution at a long distance. The system also features fast tuning, excellent stability, and low cost.

Optical video extensometer has been widely used in the field of strain measurement because of its features of non-contact, large range, high precision, being suitable for special environments and materials. A new optical video extensometer with the large field of view and high accuracy was designed in this paper. The hardware includes dual telecentric lens, CCD camera, ring light source, tripod, slide rail and some supporting fixtures. This integrated hardware has the advantages of low cost and high imaging quality. Also, the supporting software is developed. By sub-pixel tracking of selected markers, it can automatically calculate the data of transversal and longitudinal strain values. Poisson's ratio of the material can also be obtained. The measured results of the video extensometer are proven to be correct and accurate by experimental verification. Compared with strain gauges, the maximum strain error of the optical video extensometer is about 15με, and the standard deviation error is 8.38με. This optical video extensometer can be used for high-precision strain measurement of large size samples of metal, polymer, film and large deformation materials.

Camera vibration is one of the main factors affecting the measurement accuracy of stereo-DIC. It is difficult to keep fixed of the cameras in the field because of wind loads and ground vibrates. Currently, the research on camera vibration correction for Stereo-DIC in field measurements is fairly limited. This study proposes a camera vibration correction method based on fixed background points and motion model parameters. The background points are used to calculate the motion parameters of two cameras, and then the image coordinates of the interesting points that eliminate the global motion are estimated. The proposed method avoids repeated calibration of extrinsic parameters by correcting the motion of each camera separately. To verify the proposed method, an experiment in which the stereo cameras occurred relative movement was performed. Experiments shown that the proposed method has effectively corrected the effects of camera movements.

Research the imaging detection technology of tiny fiber optic endoscope based on EMCCD low-light camera for narrow lumen and low-light imaging, build a test system and conduct performance test analysis. A bimodal endoscopic imaging detection system was set up with a colour camera and a low-light EMCCD camera. The method of combining visual cognition with the definition evaluation function based on discrete cosine transform, is employed to evaluate the performance of optical fiber endoscope detection system based on EMCCD camera. Then use EMCCD low-light camera to perform weak fluorescence imaging test of pig stomach tissue . The test results show that under the given experimental conditions, the minimum light intensity is 230 Lux required for the colour camera system, while the minimum light intensity is 40 Lux required for the EMCCD camera system, and the weak fluorescence signal can be well excited and displayed under low light conditions. Based on the low-light imaging performance of EMCCD, it can be used for better endoscope miniaturization.

Structured illumination microscopy (SIM) is a powerful super-resolution method in bioscience, featuring full-field imaging and high photon efficiency. However, artifact-free super-resolution image reconstruction requires precise knowledge about the illumination parameters. In this work, we propose an efficient and robust SIM algorithm based on principal component analysis (PCA-SIM) combines iteration-free reconstruction, noise robustness, and limited computational complexity. These characteristics make PCA-SIM a promising method for high-speed, long-term, artifact-free super-resolution imaging of live cells.

Photonic Hooks (PHs) are non-evanescent light beams with a highly concentrated curved optical field. Since their discovery, PHs always have one single inflection point and thus have a hook-like structure. In this work, a new type of PHs with two inflection points and S-shaped structures were reported for the first time. We theoretically studied the effects of various physical parameters on the generation of S-shaped Photonic Hook (S-PH). The S-PH may have potential applications in super-resolution imaging, subwavelength micromachining, particle and cell manipulation, etc.

Droplet-based microfluidics is an emerging interdisciplinary field with broad applications. Accurate control of droplet volumes allows to perform quantitative detection with high throughput. But there are still many issues with the precision and practicability of droplet generation when the device suffers from inevitable disturbances or pressure oscillations. When the droplets of certain sizes are further desired, the passive system cannot generate droplets accurately and uniformly with a significant deviation between the actual volume and target. In this work, we propose the usage of Quantitative Phase Imaging (QPI) technique for real-time detection of flowing droplets, which can provide a precise 3-D measurement of droplet volume as a feedback signal. Incorporated with a closed-loop feedback control scheme to constantly monitor the instantaneous volume error and regulate fluidic pressures to keep the droplet size towards the setpoint value. The PID feedback control can develop higher accuracy and monodispersity of on-demand droplet generation, together with enhanced stability to resist internal or external perturbations, which could promise a predictable outcome in complex conditions of practical microfluidics-based systems.

As a label-free method, quantitative phase imaging (QPI) is promising for long-term observation and analysis of living cells which is crucial to biomedical research. However, a critical challenge that hinder wider implementation of this technology is the problem of hardware stability. The temperature fluctuation and the focus drift will introduce unpredictable time-varying aberrations across the long-term imaging which must be corrected timely. In this paper, we propose an adaptive aberration compensation method based on annular illumination Fourier ptychography microscopy (FPM) for robust long-term live-cell observation. Only 6 intensity images satisfying the NA-matched illumination condition are required for the correct reconstruction of both cell phase and the dynamic aberrations per frame. By an experiment of QPI on living HeLa cells for up to 51 hours, we have verified the validity and feasibility of this method.

Aiming at the requirement of remote acquisition of sound wave in transparent sound insulation screen, a method using common-path digital holography (DH) is proposed. The expanded and collimated laser illuminates on the reflector near the sound source in the sound insulation screen. The reflected light is divided into two beams in the receiving optical path. One of them is the object light, and the other is the reference light which is filtered by a spatial low-pass filter to retain the zero-order light. Then, a hologram is formed by interference of reference and object light on the surface of image sensor. A high-speed camera is used to capture the time series of hologram. The captured holograms are processed to obtain the phase distribution of the reflected light field. The remote acquisition of sound wave in the sound insulation screen is realized by selecting the phase of a certain point in the phase distribution map to form a phase time series, which represents the sound wave emitted by the sound source. Due to the common-path system, the detecting results are not affected by the airflow disturbance and environmental noise in the propagation path of the probe light. At the same time, there is no optical path length matching problem between the reference and object light. The experimental system is built with a high-speed camera with speed of 4230fps to verify the proposed approach. The experimental results show that the sound wave in the transparent sound insulation screen at a distance of 10m is successfully obtained by

Traditional topography measurements for rough objects are usually based on point-contact scan. The way of point-contact scan is usually not a high-efficient way to obtain data. Hence, in practice, only a few traces are measured which result in limited data. In order to obtain topography of sufficient amount of data, the demand for rapid noncontact measurement is increasing. In this work, we propose a dual wavelength digital holography (DWDH) measurement system with a synthetic wavelength of 51µm. We apply the DWDH system for 3D measurement of rougher surfaces and objects with a complex and small-size structure to further explore the measurement capability of DWDH. The proposed system enables the possibility to obtain the entire topography with millions of points of a rough surface in a single noncontact measurement. Firstly, we measure a cylinder workpiece of 10mm radius with a large roughness of Ra = 2.6µm which causes severe noise to measurement. The measured radius is 9.8099mm, which indicates a good accuracy. The mean standard deviation of all measured points is 1.08µm, which indicates a good repeatability and good anti-noise capability of the DWDH system. Then the proposed system is applied to measure the tooth surface of a gear of 1mm module and 18 teeth which has a complex and small-size structure. The measured surface shape agrees with the ideal gear tooth surface shape well. And the mean standard deviation of all measured points is 0.94µm, which also verifies its good repeatability. Our work presents the possibility of broader DWDH applications in industrial topography measurements.

Dual-wavelength digital holography (DH) extends the depth range of measurement, but it faces an issue that it can’t focus the whole extended depth range. In single-wavelength DH, the depth of focus of a correct reconstruction distance is usually sufficient to cover the whole depth range. Therefore, in single-wavelength DH, autofocusing algorithms which can obtain a correct reconstruction distance have been proposed and can solve the issue of focusing the whole depth range. But in dual-wavelength DH and even multi-wavelength DH, the extended depth range of the measurement often spans multiple depths of the focus of the optical system, which means that the depth of focus of a single reconstruction distance is not capable to cover the whole extended depth range, making part of depth out of focus. Therefore, in dual-wavelength DH, only autofocusing is not enough but extended focusing is demanded. However, we lack extended focusing approaches in dual-wavelength DH. Therefore, we propose an extended focusing approach based on correlation coefficient (CC) to focus the whole extended depth range in dual-wavelength DH. The proposed approach includes three steps: (1) dividing the extended depth range by the depth of focus of the system; (2) using the CC-based autofocusing approach to find the correct reconstruction distance of each divided region and reconstructing this region; (3) concatenating reconstructed image of each region and constituting the whole-depth-focused image. Through simulation experiments and practical experiments, the effectiveness and necessity of the proposed approach are verified.

Accurate characterization of physical properties of single cells is of great significance in biological research. Digital holographic microscopy (DHM) is a non-contact full-field microscopy technique that provides phase images by directly measuring the optical path differences, which facilitates cell segmentation and allows the determination of several important cellular physical features, such as dry mass. We present a series of algorithms for cell segmentation and calculation of cellular dry mass, forming a complete method for cell analysis. The image processing algorithms involved are implemented on the high-resolution artifact-free quantitative phase images obtained by our previously proposed slightly off-axis holographic system (FPDH) and associated reconstruction methods. A highly robust automated cell segmentation method is applied to extract the valid cellular region and then determine the dry mass of each individual cell. Experiments on live HeLa cells demonstrate the validity and effectiveness of the method, revealing its potential for diverse biomedical applications.

The apparent depolarization occurs when the broadband optical radiation propagates along a Spun Highly Birefringent (SHB) fiber, which will cause a decrease in polarization sensibility as well as the performance of the polarimetric sensor using the fiber. In this paper, a fiber half-wave plate made of the helically wound SHB fiber is used to compensate for the depolarization in the fiber. The half-wave plate is created by winding the SHB fiber around a cylinder and it is located at the midpoint of the whole SHB fiber. Therefore, the half-wave plate is an embedded fiber-based compensation element. The degree of polarization in the SHB fiber has been researched by simulation and experiment. The measured result suggests the excellent performance of our compensation method.

Torsion test is one of the most basic tests in material mechanics, which can reflect the static and dynamic properties of materials. Torsional deformation measurement, as a key technology in torsion test, has received more and more attention in the field of mechanical engineering. Although there have been many measurement methods for measuring torsional deformation, these methods suffer from various disadvantages. To realize the torsion measurement of large torsional deformation, this paper proposes a torsion measurement method based on the Digital Image Correlation (DIC) method, which has the advantages of full-field, large-scale, and non-contact measurement. First, to solve the de-correlation problem caused by large torsional deformation, this paper proposed an adaptive window shape based DIC method, which adaptively changes the window shape with the torsion of the specimen to ensure the correlation between the reference and deformed sub-region at a high level. Second, the monocular camera was used to photograph the specimen from six directions to obtain six reference images, and the reverse matching method was used to overcome the problem of tracking points moving out of the camera's field of view. Experiment shows that, the adaptive window shape based monocular DIC method can effectively measure the large torsional deformation.

Displacement measurement method provides quantitative information for structural safety evaluation and maintenance purposes. To overcome the difficulties in the outdoor measurement of large-span bridges, a high-resolution camera on an Unmanned Aerial Vehicle (UAV) was used to implement the deflect detection. First of all, the position of the UAV (including three displacements and three angles) was calibrated based on the speeded up robust features (SURF) algorithm. The SURF algorithm was used to extract the position information of the static areas in the image, and the calibration motion of the UAV was obtained by comparing the position changes of the static areas. The calibration used several self-designed marks with dots and circles to improve the accuracy. Then, the actual vibration of target was extracted from the raw data by subtracting the calibrated motion of the UAV. The numerical simulation indicates that the maximum error of displacement measurement can be controlled within 0.2 pixels. The calibrated experiments in the lab demonstrate that the overall correction accuracy of the displacement measurement is about 1.3 pixels. However, the maximum error of displacement measurement can be decreased to about 0.3 pixels if the rotation of UAV can be restrained by using damping mechanical devices.

The mechanical-thermal coupled load affects the stability and reliability of Printed Circuit Boards (PCBs) during their service life. The synchronous acquisition of temperature-strain data can provide supports for PCB evaluation and structural design optimization. In traditional methods, the resolution of the visible light image is higher than the infrared, which results in the low accuracy of multi-source image fusion and the difficulty in high-precision multi-field synchronous measurement. Therefore, a multi-field coupling measurement method is proposed based on the fluorescent speckles, which simultaneously measure the temperature and deformation fields. On one hand, the speckles are prepared by the fluorescent paint, whose fluorescence intensity ratio can change regularly at different temperatures for the dual-waveband temperature measurement. Then the grayscale ratio-temperature curve of the fluorescent speckles is calibrated to measure the surface temperature field based on the colorimetric method. On the other hand, the speckle images before and after the deformation are used to measure the surface strain field, which is analyzed by the digital image correlation based on integral graph acceleration. Experimental results prove that the proposed method can obtain the temperature and deformation information based on speckle grayscale ratio and correlation, and also can realize the high-precision synchronous measurement of the surface temperature and strain.

Digital volume correlation (DVC) is a powerful technique to measure the deformation inside the material or a structure. In the DVC working steps, the displacements of points of interest (POIs) are first calculated by tracking the correspondences between the volumes acquired before and after deformation using speckle correlation, which is then followed by strain calculation using the derivative of these displacements. Generally, displacement accuracy of the POIs heavily depends on the speckle quality. Unlike the speckles on the surface of test object for digital image correlation analysis can be painted or fabricated, however, the speckles inside the material or a structure for DVC analysis are often not controllable. In addition, the volume acquired by the common volumetric imaging equipment generally has limited resolution and large image noise. As a result, the poor speckle quality and large image noise introduce great displacement outliers. Strain calculation is vulnerable to displacement noise because of the high sensitivity of differential calculation to noise. Therefore, it is critical to discard the displacement outliers to guarantee the strain calculation accuracy. In this work, I proposed a ticket voting method to detect displacement outliers and then exclude them in the strain calculation. In the ticket voting method, each POI is voted by its neighboring POIs based on the residual errors in strain calculation. If this POI obtains more than half of the total tickets, it will be included in strain calculation, otherwise excluded. Both simulation and experiment results showed that the proposed method had significantly higher accuracy than the conventional method; and our DVC method can avoid the miscalculated strain concentration.

Rotating machinery plays an essential role in aerospace, energy, automotive, and other industries. A resonance phenomenon will occur if the rotational speed of the structure is close to its critical speed. Experimental analysis of the modal shapes of a rotating component at critical speeds can improve its design and performance in applications. Full-field deformation measurements of a rotating structure can be realized by three-dimension digital image correlation(3D-DIC) based on high-speed cameras. However, high-speed cameras are too expensive to apply widely. This study presents a method using 3D-DIC and a down-sampling strategy, which only uses low-cost normal-speed cameras. In order to ensure that two low-speed cameras can synchronously capture images clearly, a synchronous stroboscopic device is used. A measurement experiment was executed on a rotating disc at a critical speed to verify the availability of the method. The experiments show that the out-of-plane displacement of a rotating disc at different critical speeds corresponds to the various orders of modal shape. Compared with the non-resonance state, the out-of-plane deformation of the rotating disc was significant. The results also show the proposed method is a low-cost, non-contact experimental tool to study the dynamic behavior of a rotating component.

Three-dimensional (3D) digital image correlation (DIC) based on fluorescent speckle pattern is becoming popular for measuring 3D profiles and deformations. However, the simultaneous monitoring of both the front and rear surfaces remains challenging due to the different refractive indexes of light in different media. In this study, we utilized a multispectral 3D DIC method using fluorescent speckle patterns and a 3CCD camera for the front and rear surfaces measurement of transparent objects. Fluorescent speckle patterns excited red or blue light were sprayed on the interested surfaces before measurement. Then, the red channel and blue channel sub-systems of a virtual 3D DIC system with a 3CCD camera were used to simultaneously capture both the front and back surfaces, respectively. To unify the sub coordinate systems of different channels, a calibration model of the 3CCD camera considering the sensor designed optical path difference (OPD) was proposed. Furthermore, profile reconstruction was corrected based on Snell's law to eliminate the refractive distortion caused by the different refractive indexes. The experimental results of transparent flat and curved objects showed that the accuracy of this method was significantly improved.

Super-cavitation morphological parameter information of the cross-media projectile reflects high-speed projectiles motion characteristics. Accurate measurement of super-cavitation morphological parameters is the basis for high-speed projectile structural design and optimization, becoming a research hotspot of the underwater ballistic testing field. With the development of the high-speed photography technology, recording the evolution process of projectile cavitation morphologies by high-speed cameras and acquiring super-cavitation morphological information through image processing methods have become one of the main research directions. Aiming at reducing the large workload of manual interpretation from super-cavitation images, a multi-scale space-based automatic contour extraction algorithm for super cavitation images captured by the array cross-media high-speed imaging system is introduced in this paper. The Canny edge detection method is applied to obtain the binary edge image with coarse extraction super-cavitation edge after image pre-processing. The super-cavitation contour point set is searched in the binary edge image and small gaps between adjacent endpoints are filled at the same time. The contour extraction algorithm based on the multi-space curvature characteristics is presented to calculate the curvature value of each contour point in the high-scale space, and the curvature feature points are determined by the threshold judgment approach. The contour feature points extracted in the high-scale space are located in the low-scale space precisely, and the super-cavitation contour is extracted. Experimental results show that, the proposed method can reduce the workload from manual interpretation while improve the accuracy of contour positioning, which can provide basic data support for the cross-media high-speed projectile super-cavitation morphological parameter measurement.

Polarimetric imaging can significantly improve the quality of images captured under poor imaging conditions, these include enhancing visibility in fog, increasing image contrast under complicated background, highlighting details of targets by removing strong surface glare, etc. Compared with other polarimetric imaging systems, the full-Stokes-vector division-of-aperture polarimetric camera has obvious advantages such as the real-time imaging and the structural compactness, owing to its inherent capability of capturing the same scene’s four images related to its full-Stokes vector simultaneously. However, the deviations among four polarized images are inevitable due to assembly errors of four sub-apertures and the corresponding imaging channels. In this paper, we propose a method to solve the image registration problem in such a kind camera. Firstly, the coarse image registration is implemented through the phase-only correlation algorithm; secondly, the feature points of the reference image and other three images under registration are extracted and matched by using the Speeded-Up Robust Features (SURF) algorithm; finally, the affine transformation matrix between the reference image and each of the three images to be registered is obtained by using the RANSAC (i.e., Random Sample Consensus) optimization algorithm, respectively. The experimental results demonstrate that four polarized images are registered, which effectively enhance the image quality in terms of the Degree-of-Polarization (DoP), the Angle-of-Polarization (AoP), the structural similarity (SSIM) index, and the Normalized Mutual Information (NMI) index.

Image stitching can combine multiple adjacent images with overlapping areas into a single high-resolution panoramic image that depicts the complete information about the target scene, this technique can be well applied to the visual measurements of optical profilometry instrument. However, there may still be relatively obvious splicing traces at the edges of the overlapping areas after image stitching with traditional weighted average fusion algorithm or linear weighted gradual fading-in and fading-out fusion algorithm. To address this issue, this paper proposes an image stitching algorithm that uses the improved Sigmoid function to calculate the non-linear weights. Firstly, the extended phase correlation operation and image resampling are used for image registration, and then the weight information of the target image is calculated by using the improved Sigmoid function, in other words the original Sigmod function is scaled, so that the function is shifted right and stretched as a whole, and the domain is set as the left boundary of the overlapping area to the right boundary of the overlapping area, the range corresponding to the domain is regarded as the weight information of the target image, the weight of the reference image and the weight of the target image satisfy the relationship of summing to one. Finally, the calculated two non-linear weights are used for image stitching. The experimental results show that the improved image fusion algorithm works better whether the camera is shooting with or without exposure differences.

Light is almost exhausted in the deep sea due to the attenuation of light caused by absorption and scattering, so artificial light is often required for deep water exploration. Under the influence of artificial illumination, light scattering and absorption, the image captured by imaging system often suffer from uneven illumination and contrast reduction. To address these issues, this paper proposes an effective method for deep-sea image recovery based on multi-scale fusion. Firstly, the luminance channel of the image is decomposed into a base layer and a detail layer through Weighted Least Squares (WLS) filter, and the luminance-adjustment image is obtained by combining adaptive luminance mapping and detail enhancement techniques. Next, the model-based dehazing method is introduced to obtain a global contrast-enhanced image. Finally, two weights are designed based on the above two inputs to produce an enhanced underwater image by multi-scale fusion. Simulation and experimental results show that the method can effectively improve the details in the shadow region of deep water images and eliminating the effects of uneven illumination.

The measurement of transparent objects has always been regarded as one of the difficult problems in 3D reconstruction technology. For transparent objects conforming to the birefringence model, we propose a method for measuring the parameters of transparent objects based on multi-wavelength incident light. This paper analyzes the transparent objects where light enters the birefringence model. Because the transparent objects refract light of different wavelengths differently, when the lights with different wavelengths emitted from the same point and same direction passes through the transparent objects, the position and direction of light with different wavelength will come out differently. The position and direction of these outgoing rays are related to the refractive index and surface topography of the transparent objects. By obtaining the equations of these outgoing rays, we can derive the equations of the refractive index and surface topography of the transparent objects. In this paper, the incident light obtained by computer simulation is used as the initial data, and the simulated annealing algorithm is used to verify the model. The final results show that within the range of random error values given by the computer, the surface shape, curvature, thickness and refractive index of transparent objects with small error values can be obtained. At the same time, this paper introduces the Ceres Solver C++ nonlinear optimization library into the simulated annealing algorithm, which improves the speed and accuracy of the algorithm and greatly enhances the robustness of the algorithm.

Deep Learning is developing rapidly and has made breakthroughs in various fields. As a result, the number of parameters in deep neural networks (DNNs) is scaling about 5× rate of Moore’s Law. To meet the computing power needs of DNNs, new computing architectures have been proposed, and the electro-photonic computing system is one of them. With the physical characteristics of photons, the electro-photonic computing system has demonstrated great potential in improving the efficiency of convolution calculation. However, there are still much space for improvement in the energy efficiency of the electro-photonic computing system. And the high energy consumption led by the high-precision analog-to-digital converters (ADCs) is an important factor hindering the energy efficiency. ADCs dominate about 50% energy consumption of the entire system, and they will increase exponentially with the precision. But if the low-precision ADCs are used directly, they will destroy the computing accuracy in the convolution calculation and reduce the inference performance of DNNs. In this paper, we propose an energy-efficient quantized inference framework for electro-photonic computing system to balance the energy consumption of ADCs and the DNNs inference performance, which includes a multi-scale quantization scheme based on per-slice granularity and width variable optical matrix. The proposed framework can reduce the energy consumption of the system by using low-precision ADCs, and ensure the inference performance of the DNNs at the same time. The experiments are carried out in MobileNet-v2, ResNet50 and Vgg16 respectively. The experimental results show that the proposed quantized inference framework can reduce the precision of ADCs by 7 bits and compared with using the low-precision ADCs directly, it improves the model accuracy of 40.5%, 70.2% and 65.4% respectively, which is close to the full precision model.

Due to the influence of various factors in the production process of metal industrial products, there will be many kinds of defects on the surface. Quality inspection on the surface of metal products is particularly important to ensure the quality of industrial production. This paper based on the Inception-Resnet-V2 network to study the defect detection of steel plates. The network model is improved by adding a CBAM attention module to the original network, which increases the weight of defect-related information, reduces the interference of useless features, and then improves detection accuracy. At the same time, we use Softpool replace the traditional Maxpool pooling method to retain fine- grained feature information and fully capture more image features. Then the detection accuracy can be further improved. In this paper, the steel plate defect images are used as the dataset to train the improved network model. The improved network model can effectively identify cracks, impurities, pitting surface, oxide scale, scratches and other defects on the surface of steel plate. We compare the detection accuracy of the proposed network model with the VGG16, Resnet50 and YOLOv5 network models. The results show that the proposed network architecture significantly improves the defect detection accuracy

Optical Coherence Tomography (OCT) has been widely applied for noninvasive cross-sectional imaging in biological systems. When the incident light of OCT system meets with the particles in the internal tissues of organisms, light interference inevitably occurs, and irregular granular speckle noise will be formed on the generated scanning image. The speckle noise will cover the important subcutaneous tissue structure, so it is necessary to denoise the OCT image to reduce data loss. However, it is difficult to obtain high-quality clean samples of OCT images, so the traditional neural network denoising method has not achieved good results in this problem. This paper proposes a noise reduction method of fingertip OCT image based on generated unpaired high-quality datasets. In order to solve the problem of lacking noiseless images as truth values for network training, firstly, a small number of high-quality noiseless images are obtained by artificial enhancement. Then SinGAN network is used to train a single noiseless image to generate a large number of images with similar styles and different structures as truth values and create a dataset. Finally, CycleGAN is used to transform OCT images from noise domain to clean domain on unpaired datasets. Experiments show that this method can effectively remove speckle noise in OCT images and is superior to the traditional neural network method.

Fringe projection structured-light 3D measurement technology is an important tool in 3D digital measurement and detection in intelligent manufacturing. However, the contradiction still exists between the measurement accuracy and the measurement speed. It has great significance to introduce the U-Net network to predict wrapped phase by only using single amplitude-modulation fringe pattern. The influence of the U-Net network structure on the fringe pattern feature extraction is analyzed in detail and the structure and hyper-parameters of the classical U-Net network are improved for the phase demodulation task. It is concluded that the multi-scale symmetric structure design can provide a solution for preserving the phase information of edges and jump regions. To verify the effectiveness of the proposed method, the trained U-Net model and traditional methods are both employed to process the same fringe pattern in the contrast experiment. Then the advantages of the deep learning method, multi-step phase-shifting method and the Fast Fourier transform method (FFT) are compared. The results show that the phase demodulation accuracy of the proposed method still reached the accuracy obtained by more than three fringe patterns when the available information was only one fringe image. This method can replace the traditional FFT method in dynamic measurement applications and retain the accuracy of static measurement.

With the maturity of deep learning methods and the collection of a large number of retinal images, many deep learning models have been used for retinopathy analysis, but most of them are supervised methods. That is, the training of the models relies heavily on a large number of annotated retinal images, which is sometimes difficult to satisfy in medical fields including ophthalmology. This paper explores the application of self-supervised methods that do not require a large amount of labeled data in the analysis of retinal images. Firstly, we propose a novel self-supervised task of column dislocation and compare the performance of various self-supervised tasks on the retinopathy classification, verifying our proposed task’s effectiveness. Then, by considering the characteristics of retinal images, namely the Region Of Interest (ROI) in Spectral Domain Optical Coherence Tomography (SD-OCT) images is less than half of the whole images, we also propose a ROI based self-supervised strategy to improve the feature representation ability. The experiments on two public datasets for retinal image classification and segmentation demonstrate that the proposed strategy is effective compared with several existed self-supervised learning methods.

Processing of face images is an important branch of machine vision. In real scenarios, the quality of the acquired images often does not reach the ideal condition therefore producing wrong results. While face super-resolution and rotation are two different ways to enhance the quality of face images, these two methods have always existed independently. The super resolution of face is to improve the image quality and get high resolution image from low resolution face image, and the rotation of face is to get the image under different view of the target and enrich the target person information. Both techniques have different approaches, but both are designed to improve the image quality for easier processing such as classification, detection and recognition. Therefore, can we combine the two methods, convert an image from a low-resolution image to a high-resolution image, and then use face rotation to obtain images of other views of the target face. We can obtain a higher quality image compared with the two independent methods. In this paper, we use a supervised learning method for face super-resolution and a self-supervised face rotation method for combining experiments, and the results show the reliability of combining the two methods.

In the production process of metal castings, certain defects are easy to appear on the surface. Before the castings are put into use, it is necessary to detect whether there are defects on the surface. In this paper, the defect detection of metal casting surface is taken as the research object, and the Centernet deep learning model is used to recognize the defect of casting surface image. Centernet does not use Region Proposal Network (RPN) and Non-Maximum Suppression (NMS) in the training process. It has the advantages of simple model, good detection effect, fast running speed and easy integration and deployment. By predicting the category heat map, the length and width map of the detection frame and the center offset of the detection frame with the same image size as the input model, the defect target can be extracted. In this paper, the image dataset is expanded by cropping, rotating, changing brightness and contrast of the images. For the pit and scratch defects of metal castings, the Mean Average Precision (mAP) is higher than 0.9 for the metal casting defect image dataset, and the software integration of the model is completed.

In-line holography has been widely used in various fields because of its advantages such as simple optical path, low requirement of light source coherence and low utilization of the camera spatial bandwidth product, but it is difficult for in-line holography to restore object information from a single in-line hologram. The traditional phase-shifting algorithm requires at least three phase-shifting holograms, moreover the error caused by the intermediate phase-shifting may be accumulated and amplified. In recent years, deep learning has been widely used in the optical field due to its advantages in data analysis. Deep learning has provided new solutions for in-line holographic reconstruction, which can solve the problems that are difficult to avoid in traditional methods. In this paper, a method for generating phase-shifting holograms based on a modified Y-4net is proposed to reduce the experimental workload in data collection and the noise in phase-shifting image, which is referred to as Ps-4net. The proposed Ps-4net can generate four virtual phase-shifting fringe patterns from a single frame hologram and calculate the phase from virtual phase-shifting holograms. Simulation and experimental results show that the Ps-4net can effectively reduce the workload of data collection and the phase-shifting hologram is generated and the noise is removed at the same time.

Lensless Coaxial Digital Holographic Imaging (LCDHI) has great advantages of wide-field, high resolution and non-destructive measurement. By removing traditional lenses and utilizing a coaxial optical path, the lens-aberration can be avoided, and the imaging process is greatly simplified, which is one of the powerful tools for observing of micro components and biological cells. With the help of an open-source hardware and deep learning technology, a simple and portable experimental device based on the principle of lensless coaxial digital holography is designed and set up in this study. In order to avoid the problems of difficult data-acquisition and time-consuming training caused by supervised learning, a deep convolutional neural network (CNN) based on an auto-encoder is embedded into the Gerchberg-Saxton (GS) iterative process of LCDHI to accomplish phase retrieval. Compared with the traditional GS algorithm, more accurate amplitude and phase results can be reconstructed by the proposed low-cost device and the designed CNN through several experiments.

We present a simple and effective method for dual-wavelength phase imaging in Lensless Reflection Digital Holographic Microscopy (LRDHM). A filter is inserted between beam splitter and mirror in LRDHM. By adjusting the filter and mirror, the propagation directions of two reference beams of different wavelengths can be separated, and thus two off axis holograms with different fringe directions are simultaneously captured by a monochrome camera. Our scheme is available for real-time dual-wavelength phase imaging but requires minimum optical element and system modification for LRDHM. A series of simulations on wavefront interference analysis for using a filter is discussed in detail. We demonstrate the validity of proposed method with a step target.

Lensless light-field imaging is the process to encode the light field information of object through an optical encoder, and then recover the light field information of object through a reconstruction algorithm. In traditional lens-based light field imaging, each microlens in microlens-array corresponds to an angular sampling, and it needs to adjust the microlens-array to alter the angular resolution. In this paper, a commercial holographic diffuser is used as an optical encoder, and a flexible overlapping segmentation method of angular sampling is proposed for the Point Spread Function (PSF). The surface microstructures of holographic diffuser allow any region of it to encode and recover light information of object independently, corresponding to an angular sampling. Calibration of PSF for the lensless imaging system is done firstly, and encoded image of object is captured; then the PSF is divided into different regions as sub-PSFs, corresponding to different angular samplings; after that, light field images of object is reconstructed with corresponding sub-PSFs through reconstruction algorithm; with these light field images, digital refocusing can be achieved finally. Different from evenly segmentation of angular sampling in microlens-array, the overlapping segmentation method divides PSF into sub-PSFs while adjacent sub-PSFs overlap each other. This improves angular resolution of imaging system and ensures low error in reconstruction of light field images. Experiments show that, the overlapping segmentation method can guarantee the reconstruction accuracy of lensless light-field imaging system while flexibly adjusting and improving the angular resolution.

2D raw image of Light Field (LF) camera needs to be decoded into 4D LF data for representation and processing. In decoding, the main lens is usually modeled as a thin lens while the micro-lens array is modeled as a pinhole array. Although this model takes into account the main lens distortion, it is still difficult to accurately characterize the complex imaging relationship of the LF camera. In order to obtain more accurate 4D LF data, this paper proposes a LF camera decoding method based on a two-plane ray model. By calibrating the ray corresponding to the pixel point of the LF camera, the mapping relationship between the real object point and the image point is established to obtain a new two- plane model of LF; then by interpolating and resampling on this two-plane, the correction of 4D LF data is achieved. Compared with traditional methods, our method improves the decoding accuracy of LF camera, and provides new ideas and methods for studying LF imaging.

Fourier ptychographic microscopy (FPM) is a developed microscopic imaging technique which can break the limitation of the space-bandwidth-product and realize wide field of view, high-resolution and quantitative phase imaging. In recent years, application of deep learning in FPM effectively improved the imaging resolution of the amplitude and phase. However, these methods always require the construction of enormous datasets to train the network. To overcome the above problems, an untrained deep neural network (DNN) based on the physical model of FPM is proposed. Different from the traditional deep learning methods, this method aims to discard thousands of label data. According to the known frequency shift, the output images in DNN are virtually imaged with the forward model of FPM and a series of low-resolution images that have been simply fused are calculated to update the network by comparing errors between them and the experimental data. For a small reconstruction task, the proposed network can be treated as the iterative phase retrieval procedure, the amplitude and phase can be well retrieved with untrained parameters. The simulation results verify the feasibility of proposed physics-driven DNN. Compared with traditional deep learning methods, this method discard thousands of label data in the case of limited resolution loss.

Thermal barrier coating (TBC) system is indispensable for the heavy duty gas turbines or the aero-engine blades working at high temperature. Since the interface crack is one of the crucial issues affecting the integrity of the TBC, it is necessary to inspect the debonding defects in TBC by using a non-destructive testing (NDT) technique. Laser infrared thermography (LIRT) is a high efficiency NDT tool for thin-wall structures such like the TBC, and the LIRT using a homogeneous laser beam is especially suitable for detection of the debonding defects. In this paper, the feasibility of the homogeneous LIRT is verified for NDT of the debonding detection at first through finite element numerical simulation. Second, an experimental system using the diffractive optical element and a plano-concave lens is setup to conduct LIRT measurement with a wide homogeneous square laser spot. By using TBC specimen with debonding defects fabricated using the salt dissolving method, the pulsed LIRT and lock-In LIRT are conducted to detect the artificial debonding defects in TBC test-pieces. It is proved that the pulsed LIRT is of higher efficiency and the measured thermal image can be improved through denoising processing by using algorithm such as TSR, PCA etc. It is also found that the TSR image processing method costs less and can give results of higher contrast. The experimental results show that the debonding defects of diameter larger than 3 mm can be efficiently and accurately detected with the LIRT technique.

he safety, performance, and lifetime of lithium-ion cells are critical for the acceptance of electric vehicles (EVs), but non-destructively evaluating cell quality issues is not easy. In recent studies, batterie's deformation measurement by optical methods is a promising candidate for detecting a possible local failure originating from mechanical and current inhomogeneities or potential gradients. However, the existing measurement techniques, such as laser scanning and digital image correlation, cannot achieve high-precision deformation measurement up to submicron. In this study, a temporal electronic speckle pattern interferometry (ESPI) is proposed to acquire out-of-plane deformation of both surfaces of a commercial Li-ion polymer battery simultaneously. Various data processing techniques, such as the Fourier and windowed Fourier transform (WFT) methods, are applied in temporal and spatial domains to extract the out-of- displacement during the battery charging process. High-precision deformation of both surfaces can be obtained during the charging process. The result shows that temporal ESPI can be a helpful tool to offer high-precision data for optimizing the Li-ion batteries model.

In this paper, some research on the Laser Infrared Thermography (LIT) and the Eddy Current Infrared Thermography (ECIT) for quantitative nondestructive testing and evaluation of surface cracks and delamination of multilayer structures are introduced. At first, a Laser Array Spots Thermography (LAST) method for inspection of surface cracks in metallic structures is presented. LAST related measurement and image processing methods for inspection and quantitative evaluation of crack profiles are introduced in detail. Second, LIT technique and numerical analysis methods for detection and sizing of interface delamination of multilayer structures are introduced aiming at application to the in-vessel structures of Tokamak fusion reactors, such as the first wall panels of blanket modules. At last, some studies on the quantitative nondestructive evaluation of delamination of thermal barrier coating system of gas turbine blades with ECIT are given. The numerical methods for forward analysis of heat generation and transfer process of infrared thermography, and for inverse analysis of defect size with infrared thermography signals are emphasized in this paper, in addition with experimental validations of both the numerical methods and the new inspection techniques.

Double-sided rapid polishing machine is a new type of polishing equipment which utilizes the principle of chemical mechanical polishing and the shape precision copying mechanism to remove materials from the upper and lower surfaces of optical elements. To analyze the principle of double-sided rapid polishing machine, a material removal model based on Preston equation was established, and the relationship between the parameters such as eccentricity, rotational speed ratio and swing stroke and the material removal uniformity was obtained, which provided a theoretical basis for the double-sided rapid polishing process. With 430mm×430mm×10mm large diameter optical element as the experimental object, the double-sided rapid polishing test was carried out, and the precision index of surface shape accuracy λ/2 was obtained, which realized the high efficiency and high precision machining of plate optical element.

In order to improve the fatigue life of axial piston motor, piston holes were strengthened by inner rolling strengthening process after precision machining. Based on the principle of surface rolling, finite element simulation models of inner rolling and fatigue were established to research mechanism of residual stress and fatigue cycle, respectively. We analyzed the distribution characteristics of residual stress in hole wall and studied the influence law of different rolling process parameters on residual stress. The fatigue life of workpiece with and without residual compressive stress was compared stress life simulation. The results show that the maximum residual compressive stress of 150~210 MPa can be generated after rolling strengthening. The fatigue life was increased by about 21.4% due to residual compressive stress.

The plunger hole machining using precision boring technology has the advantages of reducing cutting force, suppressing chatter and improving the surface quality of the workpiece. In this paper, a high-precision extended boring tool and indexing turntable device are designed, and the precision boring plunger hole process experiment is carried out. The influence of different process parameters on surface roughness was studied. Roughness prediction model is established by experimental data, and the accuracy of the prediction model is verified. Adopt particle swarm optimization, constraint conditions are established. The optimization model is obtained by substituting the constraint condition into the penalty function, and the optimization results of the machining parameters meet the design requirements of the surface roughness of the plunger hole.

Integrated electro-optical modulators play an important role in fields such as broadband wireless communication and phase code radar. Electro-optical modulator based on Lithium Niobate (LN) ridge waveguide is easy to be integrated and has good electro-optical response. In this paper, a cascade modulator based on the lithium niobate thin-film structure is designed to achieve both the phase and intensity modulation. The structure parameters of the modulator are determined through collaborative simulation, then the phase modulator and the intensity modulator are simulated. Then, the phase-intensity cascade modulator is simulated to test the output light field in three working states, respectively. The experimental results show that it not only can perform phase modulation or intensity modulation of the input light wave separately, but also can adjust the phase and intensity of the light waves simultaneously.

During the optical components figuring by ion beam, the temperature of the optical component will rise quickly because of the collision between the ions and the atoms of the optical component. In order to investigate the thermal effect of the ion beam figuring, the thermal deposition model is established based on the Sigmund theory of sputtering, and the thermal power density function of the surface heat source is obtained. Moreover, the thermal power density function is also modified based on the geometric relationship between surface of optical component and ion beam for establishing the incidence of the angle of ion beam to thermal power density. On this basis, the CAE (Computer Aided Engineering method was used to simulate the progress of the BK7 glass figuring by ion beam. And the modified thermal power density function is programed as heat flow load in the CAE model. The total power of thermal deposition under the different incidence angle of ion beam was obtained by Monte Carlo method. The result shows that the modified thermal power density function is close to the experimental result in other literatures. Meanwhile, the result shows the deposition energy and the maximum stress decrease with the increase of incidence angle.

Fringe projection profilometry is widely used in manufacturing and the accuracy analysis is the key to promote this technology in engineering applications. Research analyzes influencing factors including gamma effect, intensity noise, defocus and methods are proposed to improve the measurement accuracy. However, an analytical study is difficult to perform, and the surface shape of the measuring objects influence the fringe images which needs to be considered. In this paper, raytracing algorithm and back-propagation network are used to study the relationship between the surface shape and measurement accuracy. The fringe projection profilometry system is simulated in computer using the raytracing algorithm and the light transport coefficients are measured to improve the accuracy of the camera defocus simulation. The impact of surface shape on fringe images is analyzed, and the projection and observation angles are used as the input of the network. The truth value of the surface is known in the simulation model thus the error of the coordinate can be obtained after simulation measurement and used as the output of the network. Experiment shows that, high correlation exists between the surface shape and the coordinate error.

Fringe projection profilometry is an efficient, fast and non-contact 3D measurement technique, widely used in industrial parts measurement. However, invalid phases are common when measuring step edges such as holes, ribs and steps in industrial parts which leads to outliers in reconstructed point cloud and ultimately causes reduction of valid point cloud and dimensional measurement errors. In this paper, an error compensation method on 3D measurement of step edge is proposed. 3D measurement space is firstly divided into multiple subspaces based on binocular camera system parameters. Then a calibration method is proposed to calculate the amount of compensation for each subspace and obtain parameters of edge error model. After calibration, point cloud of object with step edge is reconstructed by projecting phase-shifting structured light fringe patterns. By using Principal Component Analysis (PCA), normal estimation of point cloud is implemented. And edge feature is extracted with combination of eigenvalues variation of the covariance matrix and first- and second-order fitting based on two-dimensional projection. At the same time, the corresponding deviation amount of each edge point is solved on the basis of aforementioned edge error model. Finally, the accurate step edge is obtained after error compensation according to the normal direction and deviation amount. Experimental results show that, the method proposed can effectively improve the 3D measurement accuracy of step edge.

We developed a novel method based on geometry to reconstruct a dynamic face solely by using the conventional structured light system, which does not need extra images, hardware components, or objects. In our method, the bridge of the nose is considered the feature region because its shape remains almost constant when a human is moving or making expressions. This three-dimensional (3D) face reconstruction method consists of the following steps: (1) spatially unwrap the phase and obtain the texture image for each frame; (2) locate the nose area for each frame with a feature detection algorithm; (3) determine the fringe order in the first frame with the cues given by traditional methods and acquire the recovered result of the feature region; (4) locate the feature region for each possible fringe order in one frame by finding the pixel nearest to the camera along the depth direction in the nose area; (5) register each 3D shape of the feature region with it in the former frame by the Iterative Closest Point (ICP) algorithm; (6) determine the fringe order by the minimal Hausdorff distance between two registered 3D shapes and then construct the face with the entire absolute phase map. Experimental results indicate that our proposed approach is capable of real-time reconstruction of a dynamic face and only three phase-shifted fringes are required per frame.

In the fringe projection 3D measurement system in which the camera pixel size is much smaller than the projector pixel size, the pixel arrangement of projector will have a significant impact on the measurement results. The main reason for this problem is that the sinusoidal property of the fringes captured by the camera deteriorates. This can be improved by defocusing the projector. In this paper, the influence of projector pixel arrangement on 3D measurement error under different defocusing degree is studied by experiments. According to the actual pixel structure of DMD projector, sinusoidal fringes with zigzag and linear pixel arrangement are designed for experiments. The fringes with a period of 8 pixels are adopted for experiment, and the phase of the fringe is calculated by four-step phase-shifting algorithm. Experimental results demonstrated that the measurement error is a tendency which increases first and then decreases with increasing projector defocusing degree. When the degree of defocus is small, the phase measurement error of linear pixel array is significantly smaller than that of zigzag array. And when the degree of projector defocus increases to a certain extent, the phase measurement error of zigzag pixel array is smaller than that of linear pixel array. The experimental results are valuable for the design of high-resolution 3D measurement system.

With the continuous development of domestic and foreign manufacturing industries, there are limitations in the omnidirectional three-dimensional measurement of large parts. In order to improve the 3D measurement range, speed, and accuracy of single-line lasers, a combined measurement method based on binocular vision positioning technology is proposed. First, the circular markers are pasted around the line laser sensor, and a virtual intermediate coordinate system is constructed based on the centers of the marked points. Secondly, a feature point energy matching algorithm is used to complete the corresponding points matching of the circular markers, and the position of the circular markers is solved based on the SVD method. Third, the relative pose relationship between the virtual intermediate coordinate system and the line laser sensor coordinate system is completed based on the hand-eye calibration model. Fourth, combined with the circular markers pose and relative pose, the local point cloud data of the line laser is unified into the global coordinate system to realize point cloud splicing. The experimental results show that the average measurement error of the measurement system is less than 0.23 mm, which basically meets the requirements of general shape measurement.

Polarized angle-resolved spectroscopy is introduced to measure transparent anisotropic films with back focal plane imaging by virtue of its rich information provided at various incident and azimuthal angles. The polarized angle-resolved spectroscopy provides a conoscopic interferogram and angle-resolved spectrum to deal with the complex characterization parameters of an anisotropic film, including the thickness, principal refractive index, and optical axis. Firstly, the optics model of a transparent anisotropic film is built by superpositioning the incoherent waves from the front and rear surfaces with coherency matrices formalism when the optical length is larger than the coherent length, and the incoherent issue arises. Then the optical axis of the anisotropic sample can be determined by the conoscopic interference image with the melatope, which marks zero phase difference between the ordinary and extraordinary waves. The measured angle-resolved reflectance spectrum and anisotropic interference phase are fitted to the optics model of the film with coherency matrix formalism, utilizing the Levenberg-Marquardt algorithm to calculate the thickness and dispersion principal refractive index within a single shot. The thickness and anisotropic refractive index are determined for sapphire and polyethylene terephthalate (PET) samples with our homemade polarized angle-resolved spectrometer. The thickness measurement results show that the relative error of the proposed method is less than 1.45%.

Thin films are widely used in the semiconductor and MEMS industry. Film thickness is an important parameter to evaluate the quality of the film. Traditional measuring methods, including reflectometry, ellipsometry, etc. are considered reliable methods for measuring film thickness. Most of these methods use point-to-point measurements that requires lateral scanning in x-y directions to obtain the film thickness at different locations, which is inefficient to measure a large area and susceptible to environmental influence. Herein, a full-field film thickness measuring method based on acousto-optic tunable filter (AOTF) and reflection spectrum is proposed. The control drive frequency interval is 0.07MHz and the AOTF is driven to select narrowband wavelengths from 520 nm to 680 nm. At the same time, as driving the wavelength scanning, the camera is controlled to capture the images synchronously corresponding to each wavelength. A standard silicon wafer with known reflectivity is used to calibrate the non-uniformity intensity of the light source at the sample surface. The reflection spectrum signal at each point of the film can be reconstructed by the image sequence. The film thickness can be calculated by using the nonlinear fitting between the theoretical and experimental data. Different thickness films are measured to verify the system performance.

Computer-generated holograms (CGH) can generate the ideal wavefront required for any aspheric surface to be measured for phase compensation. It is usually used to create a cylindrical wavefront to replace the null compensator in cylindrical interferometry. Compared with the null compensator, the error components introduced by CGH are more complex, including substrate error, etching error, coding error, etc. The error calibration of the wavefront generated by the CGH can effectively improve the accuracy of the measurement results. Since CGH will generate multi-order diffraction wavefronts, there is frequency aliasing in multiple groups of fringes, so its wavefront cannot be measured directly. A physical model of the multi-order diffracted wavefronts emitted by CGH is established, and the interference of multiple sets of wavefronts is simulated numerically. To solve the frequency aliasing problem, a tilted carrier beam is introduced by using transmission holography to separate the frequency of the target wavefront from those of the other orders. Thus, the frequency of the target order is extracted, and the cylindrical wavefront is recovered according to its frequency. In this paper, the distribution of experimental fringes is consistent with the simulated fringes, which verifies the accuracy of the physical model. The simulation results show that the proposed method can theoretically separate and recover the cylindrical wavefront well, proving the method's theoretical feasibility.

Stereo digital image correlation (3D-DIC or Stereo-DIC) has advantages of high accuracy and flexibility and is widely used for 3D shape and deformation reconstruction. However, it is difficult to retrieve information of complex structure due to the severe perspective distortion in both views and area-based matching algorithm. And on the other hand, fringe projection profilometry (FPP) has pixelwise 3D shape reconstruction ability for complex structure but lacks accurate deformation and strain analyzing ability. In this work, we combine DIC with FPP to simultaneously obtain accurate 3D shape and deformation information and further perform strain analysis. First, the complete 3D shape of complex surface is reconstructed pixel by pixel using FPP. Next, the modulation of the phase-shifting fringes is extracted as texture maps to eliminate the interference of ambient light and for further DIC processing. DIC is only used to perform temporal matching for pixel-by-pixel tracking on reconstructed 3D shape. The in-plane and out-of-plane deformations are obtained simultaneously by directly comparing the complete 3D data for each corresponding pixel. Moreover, the strain in each direction is calculated by differencing the deformation data with chain rule. Experiments on the complex dynamic scene demonstrate the feasibility of the proposed method and show potential in analysis of specimen with complex structure.

The deflection response characteristics of Ionic Polymer-Metal Composite (IPMC) cantilever actuator under Alternating Current (AC) excitations are investigated in the present work. A program based on numerical calculation platform MATLAB for Digital Image Correlation (DIC) analysis with multiple Regions of Interest (ROI) is developed. The program can be used to study the micro-scale dynamic lateral movement processes of the fabricated IPMC sample with Pt electrodes, which are recorded by a digital microscope. The results indicate that the deflection amplitude increases linearly with the increase of excitation voltage and from the clamped side of the sample to the free side and has approximately a negative exponent relation with the excitation frequency. Moreover, the vibration center of vibrating IPMC sample is not fixed. The program has been proved to obtain deflections of multiple ROIs simultaneously, thus is very useful for mechanical measurements.

In this paper, we demonstrate the decoupling of complex refractive index in transport of intensity diffraction tomography (TIDT) microscopy under partially coherent illumination. Specifically, with a closed-form Tikhonov-regularized solution, this method enables the decoupling of the RI and absorption from the two through-focus intensity stacks. We compared the reconstruction performance of our proposed method with other optical diffraction tomography methods by means of numerical simulations. The results showed that our method provides RI and absorption images of 3D resolution targets with minimal artifacts. Finally, we apply the technique to imaging biological samples, e.g., the Thalassiosira diatoms, revealing the intrinsic structural diversity of the biological samples.

Single-pixel imaging (SI) methods have been widely studied in recent years. However, most existing research have illustrated SI as an attractive prospect when multipixel sensors are not preferable due to cost or technological constraints. Parallel single-pixel imaging (PSI) has been introduced recently such that SI techniques is also attractive when multipixel sensors are easily obtained and provides new possibilities to traditional challenging problems. By treating each pixel on the camera as an independent SI unit, PSI captures light transport coefficients (LTCs) in highly efficient manner, and separates direct illumination and global illumination, thereby enabling 3D reconstruction under global illumination. Localization stage is of great importance for PSI, where the approximate information for visible region is obtained. In this paper, we analyze the robustness property of PSI with respective to localization stage. Robustness property of PSI states that the accuracy of 3D reconstruction data by PSI is insensitive to errors incurred in the localization stage. Firstly, we show this property from a theoretically aspect. Then, we conduct two experiments to test this property. Lastly, we also show that satisfactory 3D reconstruction results can be obtained when only partial frequencies are captured in the localization stage for a higher efficient data capture procedure.

Fiber-Reinforced Plastics/Polymers (FRPs) are widely used in the astronautic industry as the material of spacecrafts’ structural, thermal control, and satellite payload system. Ultrafast laser processing is a non-contact, versatile or efficient tool for reduced, equal, or additive material processing, which has very recently been put into use for manufacturing spacecrafts made of fiber-reinforced polymer composites. We analyzed the fitness between ultrafast laser manufacturing technology and astronautic manufacturing demand. Special attention was paid on the progress related to ultrafast laser processing of fiber-reinforced polymers used in spacecrafts, such as cutting, milling, laser marking and surface roughening for preparation of high adhesion coatings, with carbon- FRP and glass-FRP involved. Typical machining results via traditional manufacturing techniques were demonstrated for comparisons. The study may shed light on the practical application of ultrafast laser processing of FRP products used in the astronautic industries.

We experimentally demonstrate wavelength-tunable Self-Mode-Locking (SML) operation generated in an Optically Pumped Semiconductor Disk Laser (OP-SDL) with a straight cavity. The operation is achieved by insetting an etalon into the cavity, and wavelength tuning range of 11 nm can be achieved by adjusting the angle of etalon. After aligning the cavity carefully, stable self-mode-locking is obtained when the pump power was beyond 5 W, and the pulse period of 0.99 ns agrees well with the round-trip time determined by the optical cavity length of 148 mm. Meanwhile, the RF spectrum reveals a clean peak at the fundamental repetition rate of 1.01 GHz and the signal-to-noise ratio of the RF spectrum reaches 60 dB during the whole tuning process, indicating the stability of the pulse was quite excellent. Finally, we obtained a wavelength tunable SML optically pumped semiconductor disk laser. The experimental results prove that under the condition of adding an etalon in the cavity, the OP-SDL could remain a stable operation in a wider wavelength tuning range. This research is helpful to the development of wavelength tunable self-mode-locking optically pumped semiconductor disk lasers and hopes to obtain practical applications in related fields.

Ultrashort pulse lasers are widely used in laser spectroscopy, microwave measurements, high-speed sampling, medical diagnostics, communication systems, and many other fields. This work presents a tunable self-mode-locked semiconductor disk laser, which is based on the combination of the Kerr effect in the gain chip and the soft aperture formed by the overlap of pumping and laser spot. A linear resonator with a length of 93 mm is formed by the high reflectivity distributed Bragg reflector at the bottom of the gain chip and the external output coupler with 100 mm curvature radium. The mode-locking can be started by fine-tuning the size of the pump spot and the length of the resonator, and a stable self-mode-locked output can be achieved at 980 nm wavelength. Then, a thickness of 2 mm birefringent filter is inserted into the resonator at Brewster angle, and a continuous tuning of 37 nm wavelength was obtained by rotating the filter. The repetition rate of achieved stable mode-locked pulse train at 980 nm is 1.6 GHz, the pulse width is 4.3 ps, the maximum output power is 467 mW, and the corresponding peak power is 68 W.

Beam quality is a core issue in the field of terahertz full-field imaging. In this paper, we present a terahertz spatial filter consisting of two mounted ellipsoidal silicon lenses, which have the same eccentricity but unequal size, and an opening aperture on a thin gold layer between the lenses. At the frequency of 2.52 THz, the beam transmissivity and Gaussicity of the filtering system is investigated by simulation. COMSOL Multiphysics is used to conduct two-dimensional simulation experiments, aiming to find the appropriate aperture size and gold layer thickness, depending on the wavelength of incident light wave and numerical aperture of the ellipsoidal silicon lens. The filtering system is capable of filtering the non-Gaussian beam to a nearly fundamental Gaussian beam and achieves a high transmissivity. In terahertz full-field imaging, the system can not only obtain an imaging beam with uniform intensity distribution, but also reduce the beam energy loss caused by multiple surfaces. Besides, the lens system is applicable for a wide terahertz frequency range if the wavelength dependent part is properly scaled.

Based on the diffraction propagation mechanism of terahertz waves and the basic theory of binary optics, transmission terahertz Dammann gratings is designed, fabricated, and characterized for the working wavelength of 2.52THz. The simulated annealing algorithm is used to design the phase structure of a 2D terahertz Dammann gratings which can generate 2×2 and 6×6 diffraction beams. MATLAB is used to conduct diffraction beam splitting simulation of the designed Dammann grating. The simulation results show that the 2D diffraction efficiency of the grating reaches around 60 %, and the unevenness of the split sub-spots is controlled below 1%. During the grating fabrication process, N-type high-resistivity silicon (n = 3.4175 @ 2.52 THz) is used as the grating material, and silicon nitride (Si3N4) with 200nm thickness is used as the hard mask layer. To achieve extremely high etching depths, the wet deep silicon etching process of potassium hydroxide (KOH) solution is used as the fabrication process. Various pivotal process parameters, such as the initial concentration of each component, are optimized through many experiments. The etching depth of the order of 50 μm is achieved. The grating surface flatness and step sidewall steepness satisfy the needs of use. The characterization experiment results show that the 2D diffraction efficiency of the grating is about 50 %, and the unevenness is about 2.5 %, which basically verifies the reliability of the design theory and fabrication process.

The three-dimensional (3D) Millimeter-Wave (MMW) image is a big data with redundant information that not only blurs the image but also increases the computational load on the denoising procedure. To address this problem, we propose an intensity-density associative clustering method which mainly consists of two phases. Specifically, an intensity–based clustering algorithm, e.g., K-Means, is firstly applied on the amplitude of the image data and initially achieves denoising and data compression. Then, a density-based spatial clustering algorithm, e.g., DBSCAN, is used to further extract object information. Due to the label transferring from the retained amplitude, only valid image data can forward spatial information to DBSCAN, and as a result, the computational load on DBSCAN decreases. Also, a multi-threaded parallel computing framework is developed to exploit the distributed multicore processing for its implementation. Therefore, the proposed method can be well-adapted for 3D MMW image data in a computational efficient way. Simulations and experimental results confirm the effectiveness of our method that has good efficiency on big MMW image data with respect to the noise suppression and object information extraction.

Publisher's Note: This paper, originally published on 27 January 2023, was replaced with a revised version on 12 April 2023. If you downloaded the original PDF but are unable to access the revision, please contact SPIE Digital Library Customer Service for assistance.