Face Super-resolution Using Sparse Representation with Position Weights

The image super-resolution via sparse representation considers that the patches of images can be represented by an appropriate over-complete dictionary,so sparse coefficients of low-dimensional space are mapped into high-dimension one for synthesizing high-definition patches.We propose a face super-resolution based on sparse representation with position weights method,which use the positional relationship between target patch and the sample atom as the criteria for atomic matching,to improve the accuracy of atomic basis selection and reduce the computational complexity.Experimental results demonstrate the proposed face super-resolution based on sparse representation with position weights method outperforms the traditional schemes significantly both in subjective and objective quality.