Research And Implementation Of Single Blurred Image Restoration Techniques

Image restoration is an important issue in the field of image processing, it is often happened that adverse factors cause image quality becomes poor, then restoration of blurred image has great exploration value and practical significance. How can obtain a clearer image from being restored is the main task of restoring image. This paper based on MATLAB software platform, completing the judgment and recovery of motion blurred, focal blurred and gaussian blurred image, the applicability and effectiveness of the algorithm is verified by experiment, the thesis mainly includes the following content:First of all, through reading a large number of related literature at home and abroad, comparing the advantages and disadvantages of different recovery methods, finally The author establish the fuzzy image classification before recovering it.Secondly, blurred image through spectrum processing shows the characteristics of fuzzy type related. Motion blur spectrum is made up by dark and white stripes, defocused image spectrum is made up by dark and white circle, the unique characteristic of spectrum can discern the fuzzy type, at the same time according to the distance between the light and dark stripes and the angle estimate fuzzy kernel parameters to get the point spread function.For no apparent spectral characteristics of blurred image, thought them as a gaussian blur, and applying iterative blind deconvolution algorithm to image restoration, and the fuzzy kernel and fuzzy image’s estimation simultaneously.Finally, for determined the fuzzy type of motion blur and defocused image, this paper combines noise mean square value and the information entropy with Wiener filtering algorithm to recover, making full use of prior knowledge, the restoration results is more accurate close to clear images; adding iteration threshold segmentation domain of IBD recovery algorithm was used with correction algorithm convergence domain to gaussian blur. In the last, through the experiment of subjective and objective evaluation on the restored image, verify the validity and feasibility of the improved algorithm.