Multi-view Distribute Video Coding Via DCS In WMSN

With the advantages of wide range in rapid networking, rich sensory information, andunattended operation, WMSN has a broad application prospects in national security,environment, intelligent transportation etc. However, along with great amount of data inWMSN transmission and processing, the energy consumption is sensitive and the computingresources are limited. Thus, suitable video compression algorithm has become a researchhotspot in WMSN. Recently, the CS (Compressed Sensing) has improved new solution for thevideo compression, which has several advantages such as low encoding complexity,competitive compression performance. Based on CS, DCS (Distributed Compressed Sensing)has developed with the ability to reconstruct multiple data simultaneously by effectivelyutilizing spatial-temporal correlation among WMSN sensing data.Therefore, the paper firstly analyzed the characteristic and requirement of WMSN videocoding, and secondly proposed the multi-view video coding algorithm via DCS in WMSN.The experiment results show that the proposed algorithm can ensure the quality of image,decrease the amount of coding data and the complexity of coding end, which is beneficial toextend the network life cycle of WMSN.In this paper, firstly the concept, characteristics, application and key technology ofWMSN is introduced in detail. Secondly, some video coding methods, especially distributedvideo coding, CS and DCS are profoundly elaborated. And then, in the framework of DCS, aresidual DCVS(Distributed compressive video sensing) coding method is presented toimprove the performance. The comparison of the reconstruction performance between DCVSand residual DCVS is shown in experiments.Eventually, residual distributed video decoding model based on single view video tomulti-view video is proposed. In line with spatial-temporal correlation of WMSN sensing data,the spatial-temporal correlation DCS model based on the joint sparse model in WMSN is built.In the proposed framework, CS coding is employed to the key frame, while joint sparse andsensing to the non-key frames. The complexity and data amount can be decreased by exploitthe sparsity of residual frames. In terms of the decoding, the video sequence can bereconstructed by combining the side information of residual frames and the correspondingerror compensation method. The experiment results show that this method can reduce thecomplexity of coding end and sampling rate, also improve the WMSN image quality,contributing to ensuring the validity and reliability of the WMSN video monitoring.