A Video Compression Method Based On Dynamic Background Construction

Digital video compression coding technique is the key problem of multimedia communication. The international standards on video compression satisfied the requirements for middle or high bit rate communication, which has turned into practicality. However, for the limitation of the existing capacity for transfer and store, low bit rate communication is required more. So, how to compress more efficiently is still a hot spot.This paper researched the video compression algorithm under low bit rate. According to the relativity of the intra-frame in video sequence, we proposed a video compression method based on dynamic background construction. This method according to human vision characteristic, compress the object we cared high fidelity, and for the background which we don't care, we ignore the background change, and reconstruction the background by the construction panorama by means of dynamic background construction.The related work include the following: First, segment the video shot, makes the video in every shot construct a panorama background. Secondly, remove the targets by the Gaussian model in the video sect in the same video, get the background image sequence, and then stitch the sequence to a panoramic background, after that, get the moving object region from each frame and compress the moving object by fractal coding, For the background we just need to send the coordinates of background parts in the panoramic background. So, the volume of data is greatly reduced and the compression ratio is improved.We compress the object region by fractal coding and construct the panoramic for the background, If the fractal iteration errors and the changes of the background in the same shot are neglectable, this algorithm can be done with high quality images.The experimental results indicate that this video compression algorithm has high compression ratio and high reconstructed quality under low bit rate.