Research On Motion Deblurring Algorithm Of Image

With the development of computer technology,image recognition plays a more and more important role in people's life,such as the recognition of car license plate,fingerprint identification of mobile software and face recognition of security system.Obtaining clear images is an important prerequisite for these image recognition techniques.Because motion blurring is one of the most influential factors that affect the quality of images,image deblurring is one of the key technologies of image recognition technology.Image deblurring is the inverse process of image degradation.In order to obtain accurate image information,the most important task is to establish a model of image degradation that can accurately reflect the mechanism of image degradation.The trajectory of blurred image motion is nonuniform and nonlinear,and the motion can be decomposed into a piecewise uniform and linear motion.therefore,this paper mainly studies the restoration of uniform and linear motion blurred images.First,a degradation model of a uniform and linear motion blurred image is established in this paper,and the estimation method of parameter of point spread function is analyzed.Four classical restoration algorithms of motion blurred images are summarized and compared.It is concluded that the Richardson-Lucy algorithm is more efficient and superior in the recovery of uniform linear motion blurred image.Then the kind of noise is analyzed and the characteristics of uniform noise are introduced.Finally,several algorithms for evaluating the image quality are introduced.This paper focuses on the Laplacian Pyramid algorithm based on CNR image.The most important feature of the algorithm is that it can reduce the noise while preserving the details of image.In order to get lower noise and better visual effect of image,,the paper combine the Laplacian Pyramid algorithm and Richardson-Lucy algorithm to process motion-blurring images.First,Laplacian Pyramid algorithm is used to preprocess the original image.Then,the Richardson-Lucy algorithm is used to remove the motion blurring of the preprocessed image.Finally,the mean filter and Laplacian Pyramid algorithm are used to denoise the image processed by the RL algorithm and a better image will be obtained after motion blur and noise reduction which could be used on the further study.It is proved by the final simulation that the improved algorithm has achieved better results both from the image definition and the subjective vision.