Blurred Image Restoration with Total Observation Error Bound

Blurred image restoration is a research focus in the field of digital image processing, Total Variation (TV) regularization can well preserve the details of image, however, the optimal regularization parameter must be taken into consideration in the traditional TV image restoration model during image restoration. As a result, a kind of blurred image restoration model is presented, which consists of a family of different regularization factors and a total observation error bound. The de-blurring and de-noising processes are involved to solve this model. In the de-blurring process, the conjugate gradient method is used to obtain an initial restored image which satisfies the total observation error bound constraint. And in the de-noising process, firstly, the restored image of the de-blurring process is set as an initial estimation, secondly, according to Majoriziation-Minimization algorithm, the minimization problem is divided into a series of simple sub-problems, finally, by minimizing these sub-problems, the eventual restored image is attained. Experimental results show that the algorithm is significant for blurred image restoration.