Motion-based localized super resolution technique for low resolution video enhancement

Abstract

Super resolution (SR) can be used for enhancing the resolution of images or video sequences. However, because of the computational complexity, SR is not an efficient way to improve the quality of a given video sequence. In order to overcome this problem we proposed a method, where we super resolve consecutive frames in a given video sequence in such a way that only the moving regions in the frames instead of the whole frames are processed, saving the computational cost. The proposed motion-based localized SR method not only reduces the processing time but also increases the quality of the resulting super resolved frames. The results based on PSNR measures show that the proposed localized method, in comparison with the global SR method, show improvement in quality and motion localization reduces the number of pixels to be registered, hence provides faster processing. copyright by EURASIP.