A 2D Observation Model-Based Algorithm for Blind Single Image Super-Resolution Reconstruction

In essence, image super-resolution refers to the transformation from small size image to large size image, that is, the increase of pixel density of image can provide more detailed information. It’s well-known that 1D super-resolution model can not be written directly into the form of 2D model, because the matrix dimension of high-solution image and low-solution image does not agree. The proposed 2D-based blind super-resolution algorithm combining with sparse representation model and TV term. The proposed method is to reduce the complexity of the operation by decomposing the blur matrix and the sampling matrix in the horizontal (row) and vertical (column) directions. The experimental results show that the proposed method can better protect the edge and provide more texture structure.