Research Of Image Enhancement Algorithm In Intelligent Video Monitoring System

In recent years, with the development of multimedia technology, the technology of video surveillance system plays a more and more important role in traffic, security, astronomical observation, industrial control and other fields. Because of the limitations of video acquisition equipment, the image resolution is too low. When the object of the video is moving or the video acquisition equipment is shaking, the image is blurred. When the signal of the image is transmitted, there will be noise mixed. Because of the three reasons, we can’t observe the details of the image. So the image restoration technology becomes more and more important.This thesis mainly studies the single image de-blurring and super-resolution algorithm. In recent years, there are many researches on compressed sensing and it has obtained a very good effect. Sparse coding and dictionary learning are two important aspects of compressed sensing. This paper is based on the Nonlocal Centralized Sparse Representation for Image restoration algorithm. I improve a similarity feature block rotation matching algorithm based on nonlocal means. Experiments show that the estimation error of proposed algorithm is much less than the traditional nonlocal means algorithm. So the sparse code noise can be reduced and the NCSR model for image restoration can get a better result.