STRUCTURE IDENTIFICATION FOR BLURRING MATRIX IN IMAGE DEBLURRING UNDER DETERMINATE BOUNDARY CONDITIONS

For image deblurring under determinate boundary conditions (BCs), the structure of blurring matrix be known is one prerequisite. However, the only existing identification for the structure is the extrapolating, and it is only feasible for that under BCs of corner type b, not feasible for that under BCs of corner type a. This makes image deblurring under related BCs still unsolvable. For the structure under BCs of any type, we propose a new identification, based on the fact that, the blurring matrix is just the multiplication of BCs matrix and extended blurring matrix, where the structure of the former can be extrapolated from BCs definition, and that of the latter is already known. The proposed identification first analyses the structure of BCs matrix, then that of the mentioned multiplication. In this paper, we analyse the necessity for new structure identification of blurring matrix, introduce the proposed identification, and report the newly identified structures of BCs matrices and blurring matrices, under Anti-reflective BCs of corner type a (ARBC-a), Mean BCs of corner type a (MBC-a), and Repeated BCs (RPBC), respectively. Experiment results validate the identified structures. Discussion shows the proposed identification has a wider application range than its alternative.