Image blind restoration based on blur identification and quality assessment of restored image

Nowadays, most of image blind restoration algorithms suffer from the problem of being unreliable and too time-consuming due to the large amounts of iterations involved in the algorithms. Moreover, because of the artifacts induced by blind restoration process, the restored images have a worse quality than the original. All the above greatly limit the application of the existing image blind restoration algorithms to real-time video processing. To solve the problems, an improved image restoration process is proposed to reduce the image restoration time while maintaining the quality of restored images. First, a novel image blur identification index is constructed to evaluate the image sharpness. The image blur identification result will be used to determine whether the following procedures should be performed. Second, a normalized sparse regularization blind restoration algorithm is used to restore the image. At last, a novel no-reference image quality assessment algorithm with luminance, contrast, structure, sharpness and ringing metric is designed to evaluate the restoration result. Experiment results show that the proposed blur identification algorithm and the no-reference image quality assessment method are effective in improving the image restoration efficiency while ensuring a reliable output.