Research On Spatial Image Restoring Based On Coded Aperture

Defocus is a phenomenon that objects at different depths have different resolution of images due to the optical refraction,which results in image blur.Defocus is one of the difficult problems in optical imaging,and it can not be solved completely.Therefore,deblurring methods become a popular research area.Introducing coded aperture in camera can provide extra information for deblurring.But this method is more difficult.This paper researches deeply in deblurring of spatial coded images.Conventional deblurring approaches consist of blind methods and non-blind methods.Blind methods use image probability distribution to estimate both blurring kernel and sharp images,which is easily distorted by noise.Non-blind deblurring methods need captur point spread function of camera system in optical anechoic chamber,which requires elaborating equipment.This paper issues a novel approach to restore the out-of-focus of coded images,avoiding those common drawbacks.First,this paper issues a method of estimating point spread function(PSF),the coded-apertures camera is first abstracted into a single lens imaging model with a mask in front of the lens.In our model,an impulse function simulating a pointolite is first convoluted with the coded-apertures mask considering the diffraction effect.Then,the resulting coded signal is further converged by the lens according to a particular object distance(depth),which is simulated by downscaling and convolution with a pupil function.In this way,PSFs at various depths can be conveniently simulated.The experimental results verify the accuracy in PSF estimation.Second,the paper put forwards an Errors-In-Variables(EIV)introduced and Total Generalized Variation(TGV)regulation based Augmented Lagrange(AL)deblurring algorithm.Considering the inevitable difference between the estimated PSF and real PSF,we modeled the error of PSF estimation and error of image restoration by EIV method.The TGV regularization is introduced against ill-posed problem.Via AL algorithm,the deblurring method is realized.Experimental results confirmed that the proposed method improves deblurring effectively.