关键词: 超分辨率, 牛顿插值, 相关性, 单帧图像

Abstract: In order to improve the Newton interpolation algorithm, a new super-resolution algorithm of single-frame image is proposed. The Newton interpolation results are computed from multi-directions by the pixels in the original image. The reconstruction results of super-resolution image are obtained by the fusion of interpolation results according to the correlation of the pixels in the original image. The improved method can not only enhance the utilization rate of the information contained in the original image, but also decrease accumulation error of interpolation. Simulation results show that the super-resolution image obtained by the improved algorithm is fine and distinct. A lot of detail information in marginal regions can be recovered validly. The method has better key performance indicators, such as peak signal-to-noise ratio, mean square error and visual information fidelity, than conventional methods.

Key words: super-resolution, Newton interpolation, correlation, single-frame image