Approach to Optimizing Restoration of Motion Blur Images with Robust Blind Deconvolution Based on New Cepstrum and Total Variation

Abstract

Blind deconvolution algorithms for blur images have been in widespread use for a decade. However most of the methods suffer from issues like processing speed defects, robustness, ill-posed ness. In view of the current research for blur images, a fast, robust blind deconvolution algorithm based on new cepstrum and total variation is being proposed in this paper. This new method can preserve image details and textures effectively. Compared with other restoration methods for motion blurred images, this new approach is characterized with good properties that no prior knowledge is required and exhibits good robustness characteristics due to use of new cepstrum. Furthermore, this new robust cepstrum-based PSF identification method is combined with the fast robust total variation-based restoration method to form a new algorithm that produces high fidelity results in image restoration. Experiment results with both artificial blurred images and natural blurred images which verify the outstanding processing effects are also reported.