Research And Realization On Video Compression Algorithm For Multimedia Sensor Network

With the development of the information technology, the demand for multimedia information isgrowing, so the Wireless Multimedia Sensor Networks which collect and transmit multimedia dataget more attention. Multimedia data can describle the characteristics of affairs better whencompared with traditional scalar data. What is more, the capacity of multimedia data is large, so themultimedia data must be compressed. Because of the low computing power of the nodes forWireless Multimedia Sensor Networks, the traditional data compression schemes are not suit for thesensor networks. Therefore, a scheme with low complexity encoding becomes one of the hotresearch techniques of sensor networks, and the scheme is called Distributed Video Coding.This thesis has a research on the knowledge of distributed video coding, especially pays moreattention on the exploration of residual distributed video coding (RDVC), and the thesis finds anapplication limitation of RDVC: Though at the relatively stable regions of an image the residualdistributed video coding has a good performance, but if it is applied for the intense motion blocks(IMBs), the situation is bad. Therefore, this thesis proposes a new residual distributed video codingscheme, which is called DCT-IMBs-RDVC. The scheme is based on DCT domain and IMBsselection algorithm. First the frame is divided into two parts (IMBs and non-IMBs). The IMBs usetraditional entropy coding scheme while the non-IMBs use the residual video coding scheme.On the basis of the shceme of DCT-IMBs-RDVC, the thesis proposes another scheme which iscalled Pix-Domain-IMBs-RDVC. The scheme takes the Baum-Welch algorithm at the decoder,which can fully exploit the image spatial correlation. Theoretical analysis and simulation resultsshow that, the proposed progranams can reduce the complexity of the coding end and improve theoverall performance of the DVC system.