Research Of Distributed Video Coding Based On Compressed Sensing

Multimedia sensor network (MSN), as a new kind of information acquisition and processing technology, pay more attention to audio, video, images in such as mass data collection, transmission and processing, compared with the traditional sensor network. It has wide application prospects in the military, civil and commercial field. This paper mainly aimed at multimedia sensor network of distributed video decoding. In the traditional digital signal processing fields, the sampling rate must comply with the Nyquist sampling theorem to ensure data decoding recovery, such a sampling rate is not realistic in mass video data processing in the energy resources and bandwidth limited multimedia sensor network. Recently, compressive sensing theory(CS), as a new kind of sampling theory, can realize the compression of signal in the process of sampling, in other word, CS can still recovering the original signal in the high probability even below the Nyquist sampling frequency sampling, so it can greatly reduce the system sampling, storage and computing costs.Distributed video coding (DVC) technology has "independent encoding, decoding joint" characteristics, can transfer the computing complexity in multimedia sensor network coding to the end of decoding, this characteristic and compressive sensing realize the compression of the signal characteristics completely, because these two technologies have the same kind of natural commonalities. This paper combine the two kinds of technology applied to multimedia sensor network (MSN) naturally. According to the problem of the existence of the key information of the loss based in rapid scene change coding environment, in this paper, a novel distributed group video coding method to image compressive sensing, this method has improved in some achievements of the reconstruction property, but the deficiency is too fixed not flexible; To this problem, this paper continue to improve its perfect, we puts forward a dynamic adaptive image group of distributed video coding method, namely insert a adaptive dynamic group separator through dynamic threshold set threshold T based on the basic group, so as to realize the purpose of the GOP groups which the video frame contents according to the size of the degree of scene change and dynamic adjustment. The simulation results show that the dynamic dividing group method for the reconstruction of the video signal visual effect more perfect, especially to the scene of the video signal changes rapidly, enhance and improve the peak value signal-to-noise ratio (PSNR) greatly, solve the rapid scene change environment the existence of the key information of fuzzy loss. So whether video signal of the scene the speed of the rate of change, the method is effectiveness and suitability.In addition, this paper gives the GOP according to different groups of the frame WZ average peak signal-to-noise ratio of the relationship between the number of figure, provides reference basis for the basic dividing group method in the actual application of the specific group, and finally the results show that the proposed method also has a low sampling rate CS characteristics.For frame reconstruction in the video decoding end of multimedia sensor network, the traditional approach is that to construct and estimate the edge information through the edge information to assist the frame of the decoding process based on video frame and the frame of the correlation between time and space, to adopts traditional H.264/AVC decoding frame technology, because of the edge information generation algorithm is one of the key technology in the distributed video coding system, and if it estimates the better quality information SI by the side of the interpolation, the effect of WZ frame the decoding of reconstruction will be better, so this paper introduces a kind of over symmetric motion estimation method of edge information, combined with compressive sensing technology, propose a new compressive sensing of video decoding scheme based on the edge information, and creates a complete set of "group strategy coding, edge information decoding" video frame decoding scheme based on compressive sensing. Matlab simulation experiment showed the feasibility and validity to this solution, and the reconstruction of the revised Rev-GPSR recovery performance has greatly improved to that of original gradient projection algorithm (GPSR), orthogonal matching pursuit (OMP), iteration threshold (IST) and etc, its reconstruction effect can effectively improve the distributed video compression rate distortion characteristics, and the set of scheme also can be used in the code of circumstances for a low sampling rate.To sum up, in this paper, we put forward distributed video compression coding and decoding method in the multimedia sensor networks, and set up a whole perfect set of distributed video compression coding and decoding process based on compressive sensing, both of them are high value of application and promotion prospects.