A Novel method for Unified Blind motion Deblurring of single / multi image / video using blur Deconvolution

Abstract

A unified blind method for single/multi image blur deconvolution and super-resolution of the images or videos. This paper proposes a method based on alternating minimization (AM) with respect unknown high-resolution (HR). Edge-emphasizing smoothing operation supports blur estimation process, to enhance strong soft edges and filtering out weak structures which improves the quality of blur estimation. In every iteration, parameters are updated, so that all salient edges are used to increase blur estimation. Filtering domain is used rather than pixel domain to improve blur estimation. Processing time of super-resolution is comparatively increased by separating upsampling and registration process.