Research On Image Blind Deconvolution Algorithm Based On Significant Component Extraction

With the rapid development of information technology in recent years,image shooting has gradually become popular and routine,and the fields of traffic monitoring,intelligent security,and automatic inspection have also developed rapidly.Image deblurring technology has attracted more and more attention and needs.Motion image deblurring is a relatively important problem in image deblurring.This article mainly focuses on the uniform motion deblurring of a single image.According to whether the blur kernel is known,the problem can be divided into image non-blind deconvolution and image blind deconvolution.Since blur kernels are often difficult to obtain in life,blind image deconvolution is more general.However,this problem is ill-posed and is a type of problem that is difficult to solve.It is necessary to extract the significant components in the image and add effective prior terms to constrain the solution process,so that the calculation process of the model can converge to a more ideal result.In this paper,the following research results are obtained by studying and analyzing the motion image blur principle and blur process.Since the priors of clear images and blur kernels are very important in the blind image deconvolution algorithm,it is necessary to seek effective priors to constrain the estimation process of clear images and blur kernels.Based on the Extreme Channel algorithm,this paper uses the convolution spectrum as the blur kernel prior,and designs a blind deconvolution algorithm model based on the Extreme Channel and the convolution spectrum.The convolution spectrum can effectively constrain the estimation process of the blur kernel,which can make the blur kernel converge to a more ideal result,and obtain a more accurate and continuous blur kernel.Finally,the blur kernel and the blurred image can be obtained by using a non-blind deconvolution algorithm to obtain the final clear image.The robustness of the algorithm is improved.The experimental results show that this algorithm has a relatively obvious inhibitory effect on the ringing phenomenon,and can obtain competitive results compared with other algorithms.In the course of the experiment,it was found that the range value of the local pixel block channel of the image was obviously reduced during the blurring process.Therefore,this paper proves it by mathematical formula derivation and data statistical analysis.Based on this feature,this paper proposes a blind deconvolution algorithm based on the local range channel prior,using the L1 norm to constrain the local range channel term,and constructs a new blind deconvolution algorithm model.In the model solving process,the sub-problems are solved separately through control variables and alternate iterations.When solving the clear image,the solution of the clear image can be quickly obtained by using the semi-quadratic splitting method and fast Fourier transform.Experiments have proved that compared with other algorithms,this algorithm can obtain a better deblurring effect.