Research On Single Blurred Image Deblurring Algorithm Based On Space Invariant

With the development of the computer vision,the demand of clear images which have high quality in all areas is higher,but in many practical scenarios,such as monitoring,measurement,the image can not be acquired repeatedly,and it is easy to be affected by various factors in the acquisition process,resulting in the decline of image quality.Blur is one of the important factors leading to image degradation.In order to meet the needs of practical applications,it is of great significance to develop effective deblurring algorithm in the field of computer vision without changing the hardware system.In this paper,a single spatially invariant blur image is taken as the research object.In order to improve the restoration quality,the main work of this paper is as follows:(1)According to the different classification of deblurring algorithm,this paper studies the research status of deblurring algorithms at home and abroad,analyzes the characteristics and degradation process of blurred image,and carries on the experiment,analysis and comparison of several classic deblurring algorithms.(2)The edge distribution of natural images has the characteristics of heavy tail distribution,which can be better approximated by the hyper-Laplacian distribution.In addition,the pixel values between adjacent pixels of natural images generally change continuously and are dominated by low-frequency components.The gradient statistics of natural images have more zero values observed from the gradient histogram.Based on the above characteristics,in order to improve the quality of the restored image,more to keep the details of image edge,etc,this paper designs a joint of Laplacian priori and sparse prior not blind to the fuzzy algorithm,the experimental results show that the algorithm in this paper to restore the clear image with good visual effect,and algorithm of evaluation index is better than that of the contrast.(3)In the process of image blur degradation,the minimum pixel value of its local image block will become larger and the maximum pixel value will become smaller.This paper explains the reason of this phenomenon theoretically and introduces it into the blind deblurring model as constraint information.The experimental results show that the algorithm designed in this paper can effectively estimate the fuzzy kernel with higher precision,and the recovered clear image has higher PSNR value and consumes less time.(4)Aiming at the problem that the deblurring algorithm is sensitive to noise,we presents an improved denoising algorithm based on non-local mean and wavelet threshold.The integration image is used to accelerate the traditional non-local mean filtering,which improves the time efficiency of calculating the similarity between neighborhoods.Then,the difference between the original image and the filtered image is decomposes and reconstructed in the wavelet domain,and the final denoised image is obtained by adding the filtered image.The experimental results show that this method can effectively retain the details of the image while denoising in a relatively short time.It is used as a preprocessing step of the deblurring algorithm to deblurring the fuzzy image containing noise,which can improve the anti-interference of the deblurring algorithm to noise.