Study On Parameter Estimation And Algorithm Of Motion Blurred Image Restoration

Image restoration is the process and method to obtain clear images from degraded images. Motion-blurred image degradation is the most common form of blur, restoration of the motion-blurred image is also an important branch of image processing. Image degradation or blurring could be caused by many reasons; in this case the person taking the image was bumped. Motion blurred images are generated in the process of taking images when the camera equipment was moved or shook unexpectedly, the end product is a blurry image. Motion blurred images exist widely and bring a lot of inconvenience to our daily lives, once certain moments thought captured through photography are lost. It also is one of the hottest researched areas globally, how can we fix all of these blurry images? From 1976, many researchers have proposed algorithm to estimate linear movement-blurred parameters. These algorithms have different characteristics, time complexity, accuracy and robustness in noisy environments.The purpose of image restoration is to get the original clear image as far as possible according the point spread function (PSF). Such images are highly dependent on the estimation of the motion-blurred parameter; the author chooses the appropriate image restoration algorithm to recover motion-blurred images after obtaining the parameters. Through the study of the point spread function (PSF) of motion blurred images, researcher attempted to remove the noise affects on the autocorrelation function by using self-adaptive wave filtering method. Motion blur angle is estimated with Hough transform based on the linear analysis, and the length of the motion blur is estimated with Fourier transform and Fourier inverse transform, those calculated parameters are ready. After getting the movement directionΦand the movements length L, choosing the appropriate image restoration algorithm-Wiener filter to recover the motion-blurred image.This study also compared other image restoration methods such as inverse filter, Wiener filtering, Lucy-Richardson and partial differential equation. By analyzing each principle and each experimental test, this study provided the advantages and limitations for each method and illustrated why Wiener filter method was superior. In the sense of statistical average, Wiener filtering method gives the best recovery in a Hierarchy of images, which has automatic inhibition effect of noise amplification, and the noise is stronger, the more obvious effects. In this case, Wiener filtering method avoids appearing noise amplification in the inverse filter excessive. Therefore, I choice the method of Wiener filtering for image restoration of motion blurred images finally. In this paper, based on the image preprocessing, parameter estimation and the restoration algorithm, an image restoration algorithm based on a frequency domain and space domain is proposed. The core of the algorithm is motion blur parameter estimation, in the estimation of the motion blur parameters, using Wiener filter method for restoration of motion blurred images, and simulation of the algorithm in the Mat lab platform and gives experimental results and result analysis. In the simulation experiments on Mat lab platform, the ideal images and the real images are used and both illustrated strongly positive results. The experiment shows that the advantages of this method are less calculation and high precision. It also has clarity and accuracy in the noisy images.This study is focused on the basic theory and key technologies of Motion blurred image restoration. However there are a lot of limitations, and this subject needs further research to discover new methodologies in more complicated circumstances. The motion blurred image restoration has a very important research value which can be widely applied to human life by restoring all of those distorted images we have.