Research On Superpixel Segmentation Algorithm Based Image Blind Deblurring Methods

Images,as one of the main sources of information,influence the production and life of human society.During image capture,it is common for imaging equipment to be affected by external environmental influences,such as physical devices and the relative motion between the photographer and the subject,resulting in distortion of the resulting image and affecting its subsequent use.Image deblurring is an important method to improve the visual quality of blurred images,and in recent years,has become one of the hot research directions in the field of computer vision,and is widely used in real life.Image deblurring can be divided into non-blind deblurring and blind deblurring,Since the blurred kernel information corresponding to the fuzzy image cannot be known in advance,the practical application of the image blind deblurring method is more valuable.To address the problem of the difficulty of blurred kernel acquisition,this paper presents an in-depth study on the estimation of blurred kernel and latent clear images from the perspective of image degradation model based on the superpixel segmentation algorithm,and proposes an image blind de-blurring method based on superpixel segmentation algorithm.The specific research content and innovation points include the following three areas.(1)Based on the SLIC superpixel segmentation algorithm,a method is proposed that can quickly obtain information about the significant edges in blurred images.The method is able to directly obtain potentially significant edge information from images with a high degree of blur for blurred kernel estimation.(2)Based on the probabilistic model of image deblurring,a blind deblurring model of image based on superpixel segmentation is proposed,which uses the significance edge information obtained by the superpixel segmentation algorithm to alternately estimate more accurate blurred kernel and ideally clearer images in a pyramidal structure.(3)The method in this paper is applied to the deblurring of images of multiple scenes,and a large number of experiments have been carried out on blurred images of multiple scenes,and the results show that the method can be applied not only to natural images,but also to the deblurring of images of human faces?text images and dark images,and can also obtain more desirable deblurring effects.