An Efficient Video Coding Scheme Based on Ridgelet Transform

The improvement of video coding performances is depended on how to improve the coding performances of motion compensation prediction error information, which is also called Displaced Frame Difference (DFD). In this paper, a novel tree structure multiscale ridgelet dictionary based multi-atoms matching pursuit DFD coding scheme is proposed. Firstly, the DFD is sparsely decomposed by multiscale ridgelet dictionary and multi-atoms matching pursuit. Then, the decomposed coefficients are adaptively quantized and encoded. Experimental results show that the performances of the new coding scheme outperform the state of the art H.264/AVC at low bit rate in term of both PSNR and visual quality.