Study Of Computational Imaging Based On Phase Encoding Technique

Currently, the requirement for optical imaging detection is getting higher and higherin many areas. Getting high-definition and high-resolution image is an important means oftarget recognition in many fields. With the development of optical technology, especiallythe great progress of detector technology makes it possible to obtain high-quality image.However, high-quality image is accompanied by the growth of storage and processingresources, at the same time the demand of optical parameters is more and more strict. Insome specific areas, such as aerospace and online testing of production lines wherehardware and software resources is very limited, the contradiction between high-qualityimages with limited resources is particularly prominent.The mode of traditional optical imaging is to get the visible image of objective, whichneeds the optical imaging system to obtain a high-quality and visible image at the time ofsampling, image processing is only for image de-blurring. In other words, the opticalimaging system and image restoration are not as unified complete system, resulting in agreat waste of resources.Under this background, in this paper we make a comprehensive optimization bycombining the optical system with computer digital image processing. Increasing anoptical encoder device in the conventional optical imaging system, directly encoding theoptical information with the phase-encode technology, than sampling by detector, we canget high-quality and visible image after image reconstruction. Unlike traditional imagingmethods:“first sampling, post processing”, the new calculation imaging methods involvedin this paper are “first processing, then sampling, and recovering”. We consider the opticalsystem as part of signal processing system, detector sampling signal after compressionprocessing, reducing the pressure of detection resource, and we can get high-resolutionimage after image reconstruction. Key features include: (1) Reducing the data of sampling;(2) High-resolution signal reconstruction from low-resolution sampling;(3) Breaking the diffraction limit imaging.The computational imaging methods involved in this paper modulate the phaseinformation of light; the majority of conventional optical imaging systems consider less ofthe phase characteristics of optical information. Compared to the amplitude modulation,phase modulation makes less damage to the intensity of the signal, and has higher signal tonoise ratio, it is more favorable to detect weak signals. We consider coherent light imagingand incoherent light imaging to discuss two phase encoding imaging methods, which arespectral plane phase encoding imaging and phase encoding random projection imaging.Two imaging methods use optical devices for data compression to realize high-resolutionsignal recovery from low-resolution sampling. We optimize the phase encoding board tomake it easier to manufacture, and give the results of the imaging simulation.Above imaging systems can break the space-bandwidth product limitation which limits theamount of information of general imaging system, achieving high-resolution imagingrecovery from low-resolution signal sampling. The sampling ratio of spectral plane phaseencoding imaging can achieve about25%, and the sampling ratio of phase encodingrandom projection imaging can theoretically achieve about25%. But in this paper, we takeline scan mode which uses the sparsity of one dimension of images. Simulation shows that,we can get good imaging results with sampling ratio of50%for the complex images, andthe sampling ratio can be lower for images whose sparsity is large. The simulation showsthat the spatial frequency response (SFR) restored by this method has higher response ratethan the limit of diffraction of conventional imaging in the mid and high frequency withoutconsidering the noise.Numerical simulation shows the feasibility of the two imaging methods, especially thesimplified phase encoding board, which works well as the complicated one, has potentialapplications.