An Edge-Adaptive Interpolation Algorithm for Super-Resolution Reconstruction

This paper proposes an edge-adaptive interpolation algorithm for Super-resolution reconstruction. The objective is to recover high-resolution image from low-resolution image. At first, from a low-resolution image, a high-resolution image is formed by bilinear interpolation and its edges are detected. Then, the edge of the original high-resolution image is refined by two approaches: the first is based on the geometric duality between the low-resolution covariance and the high-resolution covariance while the second is based on its local structure feature. Experimental results demonstrate that the proposed algorithm outperforms three traditional linear interpolation methods to improve the interpolation effect for super-resolution reconstruction.