Research On Video Compression Algorithm For Wireless Multimedia Sensor Network

As the wireless multimedia communications technology develops rapidly, there are moreand more video applications. Huge video data are produced in these applications; videocompression can reduce the amount of data information in multimedia sensor networks. In thevideo application of wireless multimedia sensor networks, because the terminal nodes'processabilities and networks resources all limited, it needs to code in lower complexity. Traditionalvideo coding standards (such as H.26x, MPEG-x) all adopt mixed coding framework, encoderusing motion compensation,the encoding complexity is 5 to 10 times than decoding. Because ofthis, it is difficult to use traditional video coding standards on wireless multimedia sensornetworks. How to make efficient use of precious wireless multimedia sensor networks resourcesand how to compress the multimedia information efficiently have already become the hotspots ofthe current study.Based on the detailed introduction about distributed video coding, this thesis makes furtherstudy and discussion on the Wyner-Ziv video coding. Transitional Wyner-Ziv video codingscheme adopting regional unified coding, may lead to the distortion problem of decodingestimation of the intense motion region. This thesis proposes an algorithm which based on regionof interest extraction for distributed video coding. The algorithm is based on image gradient field.For the regions of the Wyner-Ziv frame which are cannot coding and decoding accurately, thealgorithm extracts such regions according to the criteria of region of interest, then compressesbased on entropy coding, however, other regions which are encoding and decoding based onLDPC. To further improve the quality of decoded image, image post-processing based on spatial,using de-block effect and de-ring effect on the decoded image. Simulation results show that: theproposed algorithm can enhance the encoding efficiency of the intense motion region; thealgorithm can also reduce the rate and improve the quality of decoded image, and decrease theenergy consumption of sensor nodes ultimately.