A Local Adaptive Structure Sparse Representation Algorithm for Image Reconstruction

This paper studies the local structure similarity sparsity model in order to overcome the shortcomings of multi-frame image super-resolution reconstruction based on sparse representation. It obtains the good estimation of the sparse coding coefficient according to the local structure similarity by introducing the sparse coding estimation error term and utilizing the numerous redundancy of image sequences. This paper, with the aim to achieve a better reconstruction effect, presents a local adaptive structure sparse representation image reconstruction algorithm, which can adaptively set the regular parameters based on the maximum a posteriori estimation. In addition, the sparse coding of the image can be updated adaptively on the basis of the local structure in the iterative process, which makes the reconstruction model generalized. Experiments show that the proposed method can preserve the edge detail and smooth the non-edge region well.