Research On Distributed Video Coding For Wireless Multimedia Sensor Networks

In recent years, along with the rapid development of the wireless multimedia communicationtechnology, disposable cameras, medical applications, multiview image surveillance and so on, thesenew video applications start to appear. In these new video applications, which have the limits ofsubstantial resources (such as CPU, Battery power, Communication bandwidth and so on).Therefore, in order to adapt to the new demand for video applications, it must needs low complexityencoder. Traditional video coding standards (such as MPEG-x or H.26x), encoder uses motionestimation and motion compensation technology, the encoder is5to10times more complex thanthe decoder. The traditional video coding system is not suitable for these new video applications.How to make use of wireless multimedia sensor networks substantial resources to deal with a hugeamount of video data that it have already become hotspots of the study.The paper studies the distributed video codec scheme suitable for multimedia sensor network,corresponding improvements for traditional mono-view and multi-view distributed video coding. Inmono-view distributed video coding based on turbo or LDPC, the Wyner-Ziv frame coding in allregions without distinction, motion estimation techniques can not accurately predict the area whichare more intense exercise. Therefore, decoder needs to request more feedback information, thus notonly increases the rate, but the decoded portion of the image is still not accurate enough, cause thedistortion of some area in the decoded image. This paper proposes a Based on DCTH (DiscreteCosine Transform Hash）distributed video coding algorithm, which can get the intense motionregion and the non-intense motion region according to the criteria of ROI. For the intense motionregion, extract the DCT low-frequency coefficients, then compresses based on entropy coding,Decoder uses decoded DCT low-frequency coefficients for bidirectional hash motion estimationwhich enables the decoder to choose between past and future reference frames for frameinterpolation to obtain the best side information. For the non-intense motion region, however,decoder generates side information by Motion Compensation Interpolation. The Wyner-Ziv frameof Multiview Distributed Video Coding coding in all regions without distinction, motion estimationtechniques can not accurately predict the area which are more intense exercise. Then, the Decodercan not accurately generate the Temporal Side Information.And the Side Information is fusioned bythe Temporal Side Information and Spatial Side Information. Based on the DCTH distributed videocoding algorithms, the paper proposed multi-view distributed video coding algorithm based on ROI macroblock distinction. Simulation results show that the improved algorithms of the mono-viewand multi-view DVC are effective.