Ghost imaging is an optical technique that produces the image of an object by correlating the total amount of light transmitted through the object with the random intensity pattern that the object is irradiated with. When the technique is used with incoherent light sources, characterized by random temporal intensity fluctuations, it requires recording a very large number of distinct realizations to obtain a faithful image reproduction. In order to significantly reduce the number of realizations, one can use pre-programmed known patterns, so-called computational ghost imaging. Recently, ghost imaging was transposed into the time-domain to image ultrafast varying waveforms. Here, we report on a novel proof-of-concept experiment of computational ghost imaging in the time domain using wavelength multiplexing. By encoding different time-varying intensity patterns onto separate wavelength channels, we can perform simultaneous measurement of multiple realizations. This allows us to perform ghost imaging in real-time, without the need of probing the time-varying object repeatedly. Specifically, we use a programmable spectral filter to encode a set of 32 Hadamard-like time-varying intensity patterns onto a broadband LED light source. An electro-optic intensity modulator driven by an electrical waveform is used to create the time-varying object to be measured. The object is then reconstructed “blindly” by correlating the time-averaged transmission of each wavelength channels with the digitized form of the time-varying Hadamard patterns that illuminate the object. The temporal resolution of the measurement is currently to 0.5 s limited by the speed at which the variable spectral filter can be manipulated.
|