The Video Blurred Image Restoration Algorithm Research

With the development of multimedia technology, videos in the form of streaming media widely exist in the Internet and they can be received and played easily at present. However, videos can be affected easily by a variety of reasons during formation, transmission and recording process so that they have decreased quality and become blurred, distortion, etc, as seriously affect us to use them.In this thesis, for the blurred video which causes by relative motion between two adjacent frames or decreased quality of a single image, we give a blurred video restoration algorithm, based on both the methods of partial differential equation image restoration and motion blurred image restoration.For the motion blurred images, whether the parameters of the point spread function (PSF) can be estimated accurately affects the result of blurred image restoration. The wrong parameters may cause the images more blurred. Traditional methods always compute the width and the angle of light stripes in the center of the corresponding spectrum based on Hough transform to estimate the PSF parameters of blurred image spectrum. Considering the shortcomings of Hough transform, in this thesis we first preprocess the spectrum of blurred image and obtain the contour of center stripe in the spectrum by the snake method, then combine with the computation methods of the direction and width of the stripes outline based on Hough transforms to estimate the PSF parameters of the motion blurred image accurately and effectively, which ensure the satisfactory result of restoration of motion blurred image.On the other hand, for the blurred image caused by image quality and other factors, we consider the partial differential equation (PDE) method. We can use the PDE method to achieve blurred image restoration efficiently, which can meet the real-time request of video images. In this thesis, based on the ROF model, we use the difference method to calculate the video blurred image and obtain satisfactory restoration results.Finally, based on our proposed methods, we design and implement a prototype system, by which we can compare the restoration image with the original blurred image.