Image restoration based on a pair of noisy and motion blurred images

Abstract

A image restoration scheme using a pair of noisy and motion blurred images is proposed. The restoration scheme combines a deconvolution algorithm with a denoising technique. The denoising technique is initially applied to reduce noises. After the blur function has been identified, the residual noise from the denoising process is estimated by using deconvolution scheme. Unlike conventional schemes using the pair of images, the proposed scheme tries to estimate an optimal restoration result in consideration of small noises in the blurred images. An iterative deconvolution algorithm with optimal regional stopping rules is proposed. The images are decomposed into several regions based on a variance of the residual noise. Optimal regional terminating points are independently assigned in order to minimize a restoration error in each region. The optimal terminating point can be decided by analyzing a relation between a discrepancy error and the restoration error. The results from simulation are given to demonstrate the performance of the proposed method.