The Research Of Motion Deblurring Algorithm

Motion blurring is an important branch in the field of image restoration and been extensively studied by researchers,as a basic research work.This technique plays a vital role in medicine,astronomy and computer vision and other image quality demanding application scenarios because it can recover the image details.Based on previous research,this paper discussed the method which used MAP to restore the motion blurred image in detail,from the theory and realization of algorithm.This paper also analyzed the priori estimation model of noise,clear image and blur kernel which used in the algorithm.And it achieved a great improvement in the result of restoration.Based on the Variation Bayesian theory,the algorithm discussed in this paper used Kullback-Leibler divergence to measure similarity between the distribution of clear image and the introduced distribution.Using the Alternating Minimization method to solve the unknown variable s which exist in the process of above,fixing other variables when solving one variable.Taking the down sampling operation of the motion blurred image to constraint the direction of blur kernel estimation.And the experiment verified that the algorithm this paper introduced can inhibit the ringing artifacts effectively and obtain a higher quality restoration result.Furthermore,by analysis the characteristics of motion blurred image under low light environment,researchers found that the light streak is similar to the blur kernel.And the algorithm is presented in this paper.Using L1 regularization method to extract blur kernel from the patches which contains light streak selected by manual.Then,handle the saturate pixels which may cause the ringing artifacts and effect the restoration result.Finally,get the deconvolution result with improved R-L algorithm to.The experiment shows the effectiveness of the algorithm this paper presented.The restored image has a higher quality,and the algorithm reduces the ringing artifacts effectively.