Research on Blurred Image Restoration Quality Evaluation Method

Accurate evaluation of image quality is always being a problem. With blurry cotton images as samples, using variance methods, average gradient method, gray prediction error statistics method and information entropy method, this research evaluated restoration quality of 4 kinds of images from the perspective of grayscale contrast, sharpness, contrast and ambiguity. The results show that the 4 image restoration algorithms all improved image quality, and Wiener algorithm improved image quality best comparing with other 3 algorithms, Reg second, Blind algorithm and K-L algorithm were worse.