A Novel Video Codec Scheme Based on Compressive Sensing

摘要

In order to solve the problem of traditional video codec high sampling rate, a novel video codec scheme is proposed based on compressive sensing. At the encoding process, the frames in video sequence are divided into groups, in each group include intra and inter frames, random measurement matrix is constructed to measure different frames. Then the measurement values are quantized, the quantization codes are transmitted. At decoding process, each frame is reconstructed using the StOMP algorithm. Experimental results show that the proposed method is superiority over the tradition video codec with keeping the same quality of the video, and it can reduce quantization number significantly, realize easily. The codec scheme is more efficiently for non-rapid moving video sequence, the PSNR is above 45 dB.