This PDF file contains the front matter associated with SPIE Proceedings Volume 11197, including the Title Page, Copyright information, Table of Contents, Author and Conference Committee lists.

Digital Electronics and Optical Navigation Camera (DE-ONC) is an edge computing node of the asteroid probe HAYABUSA2. DE-ONC was developed to provide real-time image recognition performance for optical navigation. Lightweight, low power consumption and miniaturization are realized to overcome resource restrictions. It also satisfies high reliability and safety requirements of HAYABUSA2 missions. There are static and dynamic requirements for reliability and safety. The former increases reliability by adding redundancy combining the concept of functional distribution and time-division redundancy to meet resource constraints. Functional distribution mode, standby redundancy mode and hot redundancy mode were realized with the same device configuration. The real-time performance of optical navigation exploiting image recognition functions of the unit was demonstrated through the interplanetary cruising phase, as well as touch down to and taking off from the asteroid Ryugu. DE-ONC is always required to operate in the critical operation phase. In addition to that, it must always satisfy latency requirements to complete processing within a predetermined duration and to guarantee hard real-time performance. In order to satisfy these requirements, the image processing unit of DE-ONC adopts a unified language processing system and a distributed memory model with reference to a parallel inference machine, which is a so-called the second generation artificial intelligence technology. Its image processing module integrates a radiation hardened micro-controller unit (MCU) and field programmable gate arrays (FPGAs) with the language processing system and the distributed object model. We report the evaluation result of reliability and safety with real-time performance of the unit’s architecture.

Despite the ubiquity and maturity of imaging systems in the modern world, new imaging technologies continue to emerge that allow humanity to see into new spaces. For example, a new medical capability uses a laser system that can image neural activity through the skull in real-time. At the other extreme is the Dragonfly project for NASA, which uses an imaging system on an unmanned aerial vehicle to navigate, image, and possibly find life on the Saturn moon of Titan. This talk will cover these and other new imaging technologies that promise to change how we view our world.

Environmentally sound management of plastic waste is a fundamental achievement in order to move towards circular economy and sustainable patterns of consumption and production. In this work, different emerging applications for plastic waste monitoring and recycling based on hyperspectral imaging (HSI) technology, coupled with chemometric logics, are presented and discussed. Marine microplastics, one of the main environmental concerns, can be recognized and characterized by HSI, providing at the same time information related to the particle morphological and morphometrical attributes. Furthermore, some critical issues in plastic recycling can be solved by HSI, as separation of polyolefins, identification of black polymers and recognition of biopolymers in conventional plastic recycled streams.

The rapid growth of the market for Light Detection and Ranging (lidar) instruments is being driven by applications that include autonomous vehicle perception and navigation, as well as other novel applications such as monitoring insects flying in fields and fish spawning in lakes. Each of these applications comes with unique constraints that lead to very different optical, electronic, and software subsystems. This means that it is not possible to simply build one lidar and use it for all applications. This talk will discuss some of the primary design constraints for these novel lidar applications. Example photographs and design sketches will be shown and explained.

Waveguide technology for providing compact wide-angle, low-cost HUDs for partially autonomous road vehicles with scalability to meet future HUD requirements, extending beyond safety and vehicle informatics, to fully autonomous vehicles will be presented.

The extraction of multi-dimensional information from a biological specimen in the single examination is always beneficiary for extending the in-depth understanding of the specimen. This leads to increase the demand of multimodal imaging systems in recent years. Several multimodal imaging systems have been developed according to the requirements. We have developed multimodal imaging systems based on digital holography, for retrieving simultaneously the quantitative phase and 3D fluorescence imaging of the biological specimen, enabling to extract the specimen’s structural and functional information in a single examination. We have developed the multimodal system by integrating the coherent and incoherent digital holographic microscopes for retrieving simultaneously the phase and 3D fluorescence imaging, respectively.

In this paper, a deep learning approach for Synthetic Aperture Radar (SAR) Automatic Target Recognition (ATR) is proposed. The novelty of the proposed framework stems from the fact that it is based on a transfer learning scheme, where a pre-trained Convolutional Neural Network (CNN) is employed to extract learned features in combination with a classical Support Vector Machine (SVM) for classification. The efficiency of the presented approach is validated on the MSTAR dataset, where ten target classes are used. A classification accuracy of 99.27% is achieved.

In this paper, a 532nm laser polarization underwater target detection information processing system is designed. Firstly, combined with GPS detection, the polarization characteristic data and location information of the underwater target are obtained by laser polarization technique. Then the polarization characteristic database is established on the basis of existing polarization characteristic data. Analyze and compare the measured polarization data with the existing polarization characteristic data so as to quickly complete the target identification and classification. Finally, use the acquired location and classification information to reconstruct three-dimensional model of the measured terrain and restore underwater scene vividly. The research of this subject is of great significance for the marine resources development, underwater engineering monitoring, submarine environmental monitoring and so on.

If two signals, which have a same single frequency and different phases, impinge on an antenna array simultaneously at similar angles, a large error occurs in the angle estimation when the phase of two signals is different by 0 or 180 degrees. This phenomenon might occurs for low-flying target since direct signals and surface-reflected signals are on the same range bin because of relatively small bandwidth of shipborne radar. In this paper, we show the Cramer-Rao lower bound of the angle estimation lowered by using multiple frequencies. Also, using maximum likelihood estimation, we show that mean square error of the angle estimation lowered when we use multi-frequency radar.

We demonstrate a planar-optical Bragg grating based chemical sensor device for the detection of the environmentally harmful, highly ozone-depleting-substance trichlorofluoromethane. By applying an approximately 100 nm thin layer of differently substituted cyclodextrin-derivatives on top of the sensitive region of the silica-based Bragg grating sensor, supramolecular host-guest interactions of the oligo-saccharide affinity material and the chlorofluorocarbon analyte are favored, leading to a significant increase of the sensitivity towards the airborne pollutant. The all-optical sensor thus allows a reliable online in-situ detection and measurement of trichlorofluoromethane in real-time that features a limit of detection of down to 25 ppm.

The laser diode (LD) is modulated by the injection current of triangular waveform and a photo diode (PD) is packaged in the rear of the LD. The laser reflected by the target re-enters the cavity of LD and contains the target’s displacement information. The information is carried within the laser intensity and can be pickup by the PD. We call this laser intensity as self-mixing interferometry (SMI) signal. While processing the sensing SMI signals, we should carefully determine the windowing function and reduce the effect of windowing in the FFT and IFFT process while applying the mathematical model. Simulation results show that the proposed design is able to accomplish the measurement of micro-displacement with high resolution and great accuracy.

Amplitude-modulated continuous-wave laser scanners can realize 3D measurement with high longitudinal resolution. With installation of focusing optics in the laser scanning system, the high lateral resolution can be realized, which is enhanced by the focused beam spot size <100 um. However, the depth-of-focus of the focusing optics is generally several cm. The 3D data of the defocused objects are contaminated by aliases distributed by integer times of the half cycle of the periodical modulation. Aliasing is an impairment inherent in the amplitude-modulated continuous-wave scheme. We experimentally recovered the defocused data drastically by synthesizing those aliases. The ranging area can be elongated by at least ten times with such data processing compared with the depth of focus. Our results will contribute to highprecision industrial inspection for Industry 4.0.

When a part of light emitted by a laser is back-reflected or back-scattered from an external target and re-enters the laser cavity, both the laser intensity and its wavelength can be modulated. This is so-called self-mixing effect (SME), the optical feedback interferometry (OFI) utilizes such effect in an LD developed various applications. In this paper, we use a dualcavity OFI system that operating in period one state, the laser intensity from this system exhibits an oscillation with its amplitude modulated by a traditional single cavity OFI signal. The dual-cavity OFI system has the same measurement resolution as the single cavity which is half laser wavelength. This paper developed a method to improve the resolution by using fringe subdivision. Our simulation result shows that this method can achieve subnanometer resolution.

High precision, high resolution and absolute position are the key characteristics for linear displacement measurement.This paper proposes an absolute linear displacement sensor which is suitable for harsh environment. The sensor has a primary coil and a secondary coil.The primary coil is composed of two arrays of spiral coils which are arranged orthogonally.The primary coil is supplied with 4 kHz alternating current to generate two arrays of magnetic field that travel orthogonally. Meanwhile, the spatial periods of the magnetic fields are N and (N+1). The secondary coil consists of two sets of spiral coils, and every set of spiral coils induce only one array of magnetic field.With the motion of the secondary coil, two roads of signals are induced. According to the phase comparison, the absolute position is determined.The structure and working principle of the sensor is proposed. The sensor model is simulated by finite element analysis software. A sensor prototype is fabricated to be verified by experiment. The experimental results show that the measurement range is 61.5mm with the linearity 0.54%.

Star sensors achieve very accurate attitudes for space vehicles by comparing observed stars in their field of view with the cataloged guide stars. Guide star selection is a key work during the star sensor designing. Infrared star sensor provides accuracy navigation information to air vehicles which fly near the ground in the atmosphere. According to working characteristics of the infrared star sensors, 2MASS(Two Micron All Sky Survey) star catalog is adopted as the raw star source. On the basis of analyzing its data structure, the scheme of selecting binary stars as guide stars and calculating their equivalent magnitudes and positions is discussed. The procedures to select the guide stars are described completely by processing 2MASS. The guide star catalog for infrared star sensors is established. The result proves that guide stars in the near infrared band are far more than those in the visible band in case of the same limiting magnitude. Star camera of infrared star sensors can adopt systems having smaller aperture or narrower field of view.

The ocean color remote sensing is used to retrieve the water constituents such as phytoplankton generally from satellites observations. However, it is still difficult to apply this technology for the coastal regions, due to the low spatial resolution of the sensor, the limited observation periods. To address these problems, we newly developed a compact multi-band spectral imaging sensor, which is suitable for coastal ocean color remote sensing by a drone. In this paper, we experimentally evaluated our coastal ocean color observation system in Suruga Bay in Japan. The result showed the spectral shape of observed ocean color exhibited the typical color of case II water. Finally, we briefly introduce the latest-version of our radiometer.

An approach for recovering the phase information from the detected intensity was proposed in this work. Unlike the conventional approach based on the Gerchberg-Saxton algorithm, the proposed approach recovered the phase information via an alternative technique in the realm of deep learning, the residual neural network. The database we utilized to train the network was collected by a Michelson-based interferometer, where a spatial light modulator was implemented to provide the phase modulation as the phase object. As the result, the mean absolute error of each pixel was 0.0614π.

Organic light emitting diode generally has serious non-uniformity phenomena due to the instability of organic processing, called Mura. In this paper, we propose an automatic Mura detection model to mimic the human perception and detect Mura pixel-wisely. First, we extract regions of interest from the original image with different sizes of windows, and then we verify these regions by SEMU criterion. Consequently, we implement human visual properties based on the contrast sensitivity function filtering and ModelFest matching to segment Mura regions. As the result, our approach can successfully detect Mura with various sizes and shapes, which could have a great impact on the display industry.

We demonstrate a fiber optic dynamic displacement sensing system based on an unbalanced Michelson interferometer. In this system, the phase of the reference light is modulated with triangle and stepwise waveforms. The high-frequency triangle phase modulation makes the interferometer produce a series of chevron-shaped signals. From the change in temporal width of each signal, displacement of an object is measured with high time resolution. The effect of environmentally induced instability enhanced by a relatively large path length difference is suppressed by changing the phase by the step-wise modulation. A 10-kHz vibration displacement, which had not been measured without the step-wise phase modulation, was successfully measured by using a fiber optic Michelson interferometer even with a path length difference of 1 m.

We report on the fabrication process and the application potential of epoxy-based waveguide Bragg grating sensors on low-modulus polymer substrates for low-cost environmental seismic sensing purposes. EpoCore photoresist channel waveguides are fabricated on a highly flexible TPX polymethylpentene polyolefin substrate by means of standard i-line photolithography. Bragg gratings are permanently inscribed into the strip waveguides by KrF excimer laser irradiation using commercially available phase masks. From these raw sensor chips, a linear accelerometer sensitivity of 107 pm/g in a broadband detection range of at least 0.025 – 15 g could be observed. The enhanced seismic sensing capabilities of the Bragg grating sensor scheme are demonstrated by the use of a simple spring mass system.

We present a high-speed quantum cascade detector (QCD), a mid-infrared photodetector using intersubband transition, with a frequency response of over 20 GHz. The frequency response of an uncooled QCD with the response wavelength of around 4.5 μm was evaluated with a heterodyne beat measurement using two distributed feedback single mode quantum cascade lasers, and -3 and -20 dB cut-off frequency was obtained to be 5 and 20 GHz. The maximum frequency observed in the experiment was 26 GHz with a signal-to-noise ratio of over 20 dB. These wide band mid-infrared photodetectors are expected to offer a high-speed molecular sensing, free-space optical communication, and the other applications.

Eccentric infrared photorefraction is an attractive vision screening method which is widely used for uncooperative subjects, such as infants and toddlers. Unlike conventional slope-based photorefraction, a deep neural network is used to predict refractive error in this study. Total 1216 ocular image were collected by a homemade photorefraction device, whose corresponding refractive error was measured by a commercial autorefractor device, to create a series of dataset for our deep neural network. The mean squared error of the preliminary result is ±0.9 diopter, which indicates its feasibility and can be improved with bigger database.

Optical location systems implemented on the basis of high-resolution video cameras are currently used in many areas. These are, for example, medical equipment, traffic control systems, satellite monitoring systems, preventive security systems, object recognition and classification systems, etc. For these systems, the requirements for high-resolution image processing speed of 8K, 16K and more are increasing every year. Processing of such information becomes even more difficult when providing a reading of high frequency frames from the matrix of the video camera, especially for systems operating in real time and using high-speed networks of exchange, processing and integration of information. This requires to determine a set of types of information processing procedures: masking, compression, noise-protected coding, etc., for which algorithms should be revised in multi-user and multiposition application in distributed information processing and aggregation systems. Processing of such information becomes even more difficult when providing a high frequency of reading frames from the matrix of the video camera, especially for systems operating in real time and using high-speed networks of exchange, processing and integration of information. This requires to determine a set of types of information processing procedures: masking, compression, noise-protected coding, etc., for which algorithms should be revised in case of multi-user and multiposition application in distributed information processing and aggregation systems. In this regard, the problems of development and improvement of new ways of representation, compression, storage, masking and error-correcting coding of high-resolution images with a common mathematical basis are relevant. Most information transformation procedures are based on the use of orthogonal bases, in particular orthogonal and quasiorthogonal matrices. The paper presents the results of the search and formation of such bases, the methods of synthesis of quasi-orthogonal matrices for image processing problems that meet the formulated requirements. The methods of guaranteed synthesis of matrices of symmetric, cyclic, block-cyclic and other structures of different orders, assuming economical storage and generation, are proposed. Such matrix bases, which are constantly expanding, provide developers with a wide range of algorithms to choose the most appropriate one from them. The problem of search and study of extreme quasi-orthogonal matrices has great importance for a wider range of information processing tasks, not just images. The proposed mechanisms for finding new classes of matrices allow creation and development of competitive methods of storage, presentation, compression, noise-resistant coding of data during their transmission in wireless high-speed networks of exchange, processing and aggregation. The results of the work correspond to the world level of research and have a universal character, as they can be applied in a variety of fields, including orthogonal cryptography, models of crystallography and biology, in the finite models of dynamic processes, etc.

Modern industrial production technology is developing rapidly in recent decades. More and more traditional manual productions are becoming automatic when new automatic technologies are introduced among which machine vision plays an indispensable role. Aiming at the unmanned automatic production line of special steel castings in Shanghai MengTeng automatic technology co., LTD , in this paper an algorithm to recognize and position the special steel castings autonomously is proposed based on the binocular vision theory. The technologies including the pyramid hierarchical matching and the deep neural network are described in detain. The identification and positioning of special steel castings is realized. Its procedures are described completed. Some experiments are implemented by using manipulator to grasp the steel castings and running the proposed algorithm. It proves that the proposed algorithm achieves satisfactory results.

Recently the direct TOF Lidar receivers employ cell averaging constant false alarm rate (CA-CFAR) technique to determine the threshold for the reflected pulses. CA-CFAR locally estimates the mean level of the noise from a predefined interval. The length of this interval is determined by a fixed number of reference cells. CA-CFAR with small number of reference cells fails to detect long distance objects. Conversely, setting reference cells to a large number can’t resolve closely spaced objects even at short distances. This paper presents an adaptive CA-CFAR adaptively determining the optimal number of reference cells. Simulation results show the capability of the proposed method to do both detecting long distance objects and resolving closely spaced objects at short distances.

Currently, small aircraft, including UAVs are widely used for civilian purposes to obtain high-precision maps of the terrain, determining the edge of the land-sea, assessing the state of farmland, classification of observed surfaces, environmental monitoring, and also to detect physical objects and sensors that inform about the state of man-made objects. Onboard small radar located on a small aircraft must provide high resolution as coordinate range so as azimuth coordinate when solving these tasks. However, due to their small size and, consequently, the small aperture of the airborne radar antenna, the provision of high resolution azimuth is problematic, especially in the front viewing area, where it is impossible to use SAR methods. The paper considers with the problem of front and front side view of the earth's surface by a group consisting of two onboard small-size radar station combined into a single information and telecommunication distributed system. Information is exchanged between the equipment of the respective small aircraft via a high-speed radio frequency communication channel. Such a two-position small-sized radar, with appropriate integration of information obtained by spaced sources, allows to overcome the limitations of single-position systems. The paper presents algorithms of information processing in the described two position small radar system, that significantly improves resolution in the azimuthal coordinate in the front and front side zones of review at rather small apertures of the receiving antennas. Shown the requirements for the characteristics of airborne radar and trajectories of small aircraft, allowing to achieve the specified dimensions of the resolution elements on the coordinates of the range-azimuth and, accordingly, to improve the accuracy of the coordinates of the detected physical objects.

We developed a short range Lidar system using a supercontinuum source spectrally tailored to cover the ro-vibrational transition energies of desired components of a flue gas. The system enables simultaneous remote measurements of the gas parameters, like temperature and concentration which play a key role in the performance of combustion power plants. The technique requires only one inspection window and can thus be used in combustion units with limited access. It exploits differential absorption between specific wavelength bands of the gas absorption spectrum. The transmittance of individual wavelength band is derived from the detected backscattered temporal intensity of the supercontinuum pulses. We demonstrate preliminary industrial measurement of water vapor temperature and concentration in a full scale boiler. The technique also enables 3D mapping of temperature and concentration.

To obtain maximum possible details from source imagery, we propose a fast and effective image fusion method with guided saliency(IFGS). Our method first computes the saliency maps of all source images, and then each guided saliency map is generated using Logarithmic spectrum with an adaptive threshold value. Next, the weight of each component is computed by the guided gradient (FGG) method. Finally, the coefficients of the fused image are obtained by linearly combining the weights of the source images with the coefficients of the corresponding images. Experimental results demonstrate that our method can generate high-quality fusion images with better color and more details from multi-focus and multi-exposure compared to other fusion methods.

The aim of this study is to develop a new methodology based on optical sensors able to recognize and characterize microplastics collected in the marine environment. In recent years, this pollution represents an environmental emergency with serious effects on human health and marine organisms. The study was carried out on microplastics collected in different geographical areas. Through the hyperspectral imaging (HSI) based approach, performed in the SWIR range (1000-2500 nm), a classification model based on chemometric techniques was implemented in order to identify in a rapid, non invasive and non-destructive way the polymers in the investigated samples.

We propose and characterize a novel and simple architecture of dual-frequency noise rangefinder aiming at achieving high precision and robustness with reduced hardware requirements. In the scheme, the range measurement is achieved by exploiting the differential beat-noise between two lasers. In more detail, the outputs of the two lasers are coupled together and sent to the target. The backscattered echo from the target is acquired by the echo-receiver. The two lasers’ outputs are also sent to the reference-receiver. The range measurement is achieved through the digital cross-correlation between the lasers’ beat-noise information acquired by the two receivers. A laboratory demonstration shows range accuracy and resolution of 3.6 mm and 0.8 m, respectively, well in accordance with the signal/noise-dependent theoretical boundary values.

Small, imperceptible perturbations of the input data can lead to DNNs making egregious errors during inference, such as misclassifying an image of a dog as a cat with high probability. Thus, defending against adversarial examples for deep neural networks (DNNs) is of great interest to sensing technologies and the machine learning community to ensure the security of practical systems where DNNs are used. Whereas many approaches have been explored for defending against adversarial attacks, few have made use of the full state of the entire network, opting instead to only consider the output layer and gradient information. We develop several motivated techniques that make use of the full network state, improving adversarial robustness. We provide principled motivation of our techniques via analysis of attractor dynamics, shown to occur in the highly recurrent human brain, and validate our improvements via empirical results on standard datasets and white-box attacks.

We constructed a polarization lidar system with single detector for atmospheric studies. The lidar employs a rotating polarizer in the optical part of the lidar receiver. The rotating polarizer contains a Glan Thomson prism which is used to separate the p- and s- components of incoming light. The rotation of polarizer is controlled through a stepper motor controller. The lidar contains a trigger unit, which is an integral part of lidar that controls the timing of rotation of polarizer and also the sequence of lidar signal acquisition. The integrated trigger unit is synchronized with laser Q– switch pulse and provides polarization discrimination in the incoming light with high degree angular alignment. The lidar uses an Nd:YAG pulsed laser for probing the atmosphere and is operated at low repetition rate to retrieve the coand cross polarized signal components from the single detector. Using the lidar setup, we derived the range resolved depolarization measurements of tropical cirrus at the second harmonic wavelength of Nd:YAG laser.

This study presents the ultra-rapid deployable UAV antennas for VLF, LF, MF, and HF applications. The proposed compact design consisting of the flexible radiating structure carried and positioned by the multi-rotor UAV can significantly reduce the antenna size, weight, and cost. It is basically intended for tactical communication, underground/underwater radio, and surface-wave radar applications but with simpler, lighter, and cheaper structures. Simulated and measured results substantiate that the proposed method can be employed to provide satisfactory radiation characteristics for large antenna systems. Moreover, this UAV antenna can be used to realize the dynamic polarization and pattern tuning of the antenna.

This paper demonstrated a multiwavelength fibre laser incorporating semiconductor optical amplifier in nonlinear optical loop mirror with polarisation maintaining fibre. The configuration comprised 3-dB optical coupler, semiconductor optical amplifier, and a 2-meter length of polarisation maintaining fibre. The range of semiconductor optical amplifier varied from 160mA to 180mA to determine and evaluate their performance in terms of peak power and stability. Based on the results, the power fluctuation for different wavelengths at 1555nm, 1558nm and 1561nm was less than 5dB. In addition, the output peak power increased as the semiconductor optical amplifier current increased. The highest output peak power was - 2.093dBm at 180mA with lasing wavelength of 1558nm. The stability test showed that the multiwavelength fibre laser system was stable at room temperature within 60 minutes. Furthermore, the polarisation maintaining fibre that acts as gain medium has a broader bandwidth operating in the C-band to L-band at room temperature.

Breast carcinoma has become one of the most frequently diagnosed life threatening cancer among women. Early detection of breast cancer is highly essential with the aid of non-contact imaging modalities. Recently, non-contact breast imaging methods based on fringe projection has been developed for breast surface change inspection. In this work, a non-contact digital fringe projection imaging modality that utilizes phase shifting fringe patterns for identifying surface changes has been developed for investigating breast surface changes caused by the presence of tumors. A medical grade prosthetic breast was used as the experimental subject and subjugated to breast changes through the enlargement of a round shaped silicon catheter as a tumor. The fringes were projected onto the breast surface based on the three steps phase shift fringe projection. The fringe patterns consisted of a large fringe width to enable the breast image to be confined in fewer fringe patterns. A resulting phase map was obtained where pixel coordinate marking was conducted on the phase map breast image. Each of the pixel’s coordinate was compared to identify the location of the surface changes. A range of 2-9 pixel coordinate shifts from the 0.5 – 2 cm tumor growth were obtained from the results which demonstrated the capability of using phase map analysis from digital fringe projection in identifying surface changes of the women’s breast.

The resonator fiber optic gyroscope (RFOG) senses angular rotation by measuring the frequency shift between clockwise and counterclockwise resonances inside a fiber ring resonator (FRR). Three types of polarization-maintaining FRR are studied: normal FRR, FRR with 90° polarization-axis rotated splice, and FRR employing an in-line polarizer. We establish the models to analyze the output characteristics of the three FRRs, and present the experimental results for the three FRRs. In the normal FRR, the unwanted polarization state will destroy the symmetry of the resonance curve. In the FRR with 90° polarization-axis rotated splice, the unwanted and wanted polarization state coexist, and they do not overlap. In the FRR employing an in-line polarizer, the unwanted polarization state is completely suppressed, but the finesse of the FRR is also reduced.

Polarization sensor is a device to perceive the polarization information of skylight in polarization navigation. And imaging polarization sensor is better than the multi-channel point sensor for the advantages of wide field, large data redundancy and strong robustness. But the polarization patterns of skylight in different spectrums are different, which would represent different symmetry and structural similarity to the Rayleigh model. Meanwhile, the detectors inside have different responses in different spectrums. Therefore, the spectrum selection in the process of designing an imaging polarization sensor is an important issue. In this paper, we quantitatively evaluate the spectrum for imaging polarization sensor under the different sky conditions by two indexes: structural similarity (SSIM) index and symmetry error index. The result shows that wavelength shorter than 514nm has both higher structural similarity and symmetry in the visible spectrum. In other words, the imaging polarization sensor works better in the short-band spectrum.

Applications in sensor systems utilize, for example, changes in the polarization of light in an optical fibre during temperature changes. The principle of operation is based on the specific propagation of polarized light by special optical fibres, which, due to their construction, keep the light in two axes, but the speed of their propagation and thus the instantaneous state of polarization of light change. It has been shown that by suitable methods it is possible to transform changes in light polarization into changes in light intensity, which can be detected very effectively by a photodiode. This method has been practically verified, implemented by fibre components and confirmed to be suitable for implementing a fibre sensor of thermal field disruption. The aim of this work is to point out the current possibilities of using polarized light in sensor technology. This polarization system can efficiently secure areas, the surrounding environment and monitor the changes that occur.

This paper presents the automatic target detection and tracking of marine obstacles for unmanned surface vehicles (USVs). For practical applications with a USV, the automatic detection of surrounding obstacles is a crucial capability, and marine radars have been commonly used to detect and estimate the motion of obstacles. However, their tracking performance degrades when a target is approaching at a high relative velocity due to their relatively low sampling rate. This study addresses the automatic target tracking of marine obstacles by considering time-delayed measurements provided by a marine radar. The relative position information between a USV and nearby obstacles is obtained using the radar sensor, and the obstacles’ motion including position, course, and speed is estimated using an extended Kalman filter (EKF)-based tracking filter by compensating the measurement delay. To validate the feasibility of the proposed method, a real-sea experiment was conducted using a USV and the results are presented.

Polarimetric imaging data contains rich information about surface textures, shape, and shading of objects in a scene, which can be used to discriminate objects from background. Due to spectral signatures are limited to material properties, separating man-made objects from natural scene is a difficult task in complex scene. In this paper, we present a man-made object separation technique from natural scene utilizing polarimetric data. We started by calculating different polarimetric component images based on Stokes vector measurement. After that, combining polarimetric component images into color space and single-channel conversion, a polarimetric signature image is calculated. Then, considering pixel neighborhood relationship, an incremental clustering approach is applied to group similar pixel patterns of polarimetric signature image. Finally, a morphological structuring element is applied to conduct the morphological close operation to refine the final background mask. Ground truth is generated manually. A high-performing score (Dice Similarity Coefficient (DSC)) is achieved on the final man-made object area mask which separates man-made objects well from natural scene. Future work will exploit the use of multispectral polarimetric imagery for target classification using machine learning techniques.