Study On Photon Counting Integral Imaging Based On Element Images

In low-light environment,it is necessary to obtain low-light images by means of optical detector and algorithm.Photon counting integrated imaging combines photon counting imaging and three-dimensional integrated imaging.It provides three-dimensional images under low-light environment by collecting photon counting images under multiple perspectives,which is one of the most promising imaging technologies in low light environment at present.In order to improve the quality of the reconstructed image,the researchers mainly focus on the Poisson characteristics of photon-counting images to improve the reconstruction algorithm.Few researchers focus on using element images to improve imaging quality.Therefore,in this paper,we study the photon counting integrated imaging technology based on element images.Firstly,this paper study the advantages and disadvantages of different methods of three-dimensional integral imaging technology.We find that in the process of image acquisition,element images are greatly affected by the errors of optical devices.In order to reduce this effect,we study a photon-counting integrated imaging method based on computer generated integrated imaging(CGII).The element images is collected by the camera array which is simulate d by software.Then we use the maximum likelihood estimation andBayesian estimation reconstruction algorithm to reconstruct the element image.We verify the effectiveness of the CGII method.And using the peak signal-to-noise ratio(PSNR)to analyze the relationship between the reconstructed image quality and the number of photons.Secondly,aiming at the Poisson characteristics of photon-counting images,we use the composite photon counting algorithm(CMLE)to improve the quality of element images.Then combined with Lucy-Richardson(LR)algorithm,the quality of element image is further improved.Based on CMLE and LR algorithm,we conduct an experiment to analyze the influence of iteration times on element images.And we find that the peak signal-to-noise ratio(PSNR)of element images tended to be stable after 120 iterations.On this basis,we use the maximum likelihood estimation algorithm to reconstruct.The results show that the CMLE+LR algorithm can effectively improve the reconstructed image quality,PSNR improve 1.416 dB.For different models,the PSNR of reconstructed image quality improve 1.642 dB.