Research On Deep Space Exploration Video Coding Based On Distributed Compressed Sensing

Deep space video transmission equipment has huge pressure due to video data enormous, and also the channel bandwidth is challenged to satisfy the requirement of transmission. The traditional video compression technology is unable to break through the constraints of load limited in equipment resource, so seeking a new method of the new video compression technology which has high efficiency compression, high decompression quality and low coding complexity has to be considered.Compressed sensing(CS) as new theory provides a solution for the asymmetric resource environment: compress image signals linearly on the encoding end and decompress which by nonlinear reconstruction algorithm on the decoding end. This method can transfer computing complexity from the encoding side to the decoding side, so it is conductive to the communication environment which has the serious asymmetric resource and limited computing power on the encoding side. Distributed compressed sensing(DCS) is enable to obtain the high compression efficiency and recovery performance due to considering the internal correlation of a group of signals in compressed sensing. There can be generated a new system which is called distributed compressive video sensing(DCVS) by combining the theory of DCS and distributed video coding(DVC). The new system is eligible to deep space communication.To develop the recovered key frame performance of video, we propose a new recovery method by adopting the double-density dual-tree complex wavelet transform(DDDT-CWT) as the sparse representation scheme. In addition, the structural characteristics of the DDDT-CWT coefficients are utilized as extra prior knowledge in the recovery process to further improve the recovery quality.We also propose another new method to recover non-key frame of video which is based on double sparsity dictionary: Separating different scales sub-bands of the images under wavelet domain, then utilizing K-SVD algorithm to obtain different redundant dictionaries with multi-scale properties, finally deploying trained dictionaries as the sparse representation and using gradient pursuit for sparse reconstruction(GPSR) algorithm to recover non-key frames. Furthermore, we choose DDDT-CWT replaceing the traditional discrete wavelet transform(DWT) which is limited by poor directionality and lack phase space information. This new method can analyze images more accurately and therefore improve the performance of the communication system.Extensive simulation results have been conducted and the results show that under the same compression ratio, the proposed method achieves considering gain compared to the traditional recovery algorithm.