Edge Information Retained Super-resolution Restoration for Still Image
保留边缘信息的静态数字图象超分辨率重建

Interpolation is an important method for super resolution technology. We believe that visual quality of an interpolated image mainly depends on two factors of an edge profile: (1) The gray value across the edge orientation is sharp; (2) The gray value along the edge orientation is smooth. For this reason, we propose a new interpolation method in this paper. Firstly, we detect the edge using Canny operator and interpolate along the edge. Secondly, we retain edge information of the original image by retaining pixels abutting against the edge in low resolution image abut against the edge in high resolution image and we smooth along pixels abutting against the edge in high resolution image for protecting edge information. Thirdly, we process the smooth region. Finally, we interpolate pixels between the pixel abutting against the edge and smooth region. By this method, we can create a clear edge image and therefore obtain a better magnified digital image.