On The Design And Implementation Of The Sub-Block Compressed Sensing Based Image And Video Transmission

On the Internet, multimedia and video-centric applications have become the main applications. In the conventional video transmission system, the complexity of the encoding system is higher than the decoding system. Therefore, the conventional video encoding scheme is suitable for Internet environment, where the video content will be encoded once but may be decoded multiple times according to different requests. Instead, only little information is meaningful for the general video applications in surveillance networks and sensor networks. The complexity of the encoding system will be reduced with compressed sensing and the distributed video transmission technology. The light weighted video encoding system becomes much more attractive for its capability in realizing long time processing, cost-effective and energy-efficient. Therefore it has very high research value.Firstly, this paper introduces the basic theory of compressed sensing and discusses several key issues in this part. Then a comparison of several different measurement matrix and reconstruction algorithm is make by the simulation. The results show that the Gauss orthogonal random matrix which works as the measurement matrix performs well on the stability and reconstruction. The OMP algorithm also converges quickly and reconstructs well. Then the paper introduces a series of algorithms based on blocked compressed sensing, BCS-SPL. The PSNR and time cost of reconstruction in different orthogonal basis are compared. BCS-SPL-DDWT can get good results under normal conditions. The paper analyzes the effects of reconstruction algorithm with different block size. The best performance can be achieved in suitable block size. MH-BCS-SPL can improve the performance of BCS-SPL. Further experiments about color pictures verify the effectiveness of the algorithms. At last, the transport framework is compared with Softcast in the wireless multi-user channel conditions.The paper summarizes the characteristics of conventional and distributed video transmission system. The paper puts the image frames into odd frames and even frames. The even frames are considered to be the initial predictive frames of the odd frames. Then, the transport framework based on blocked compressed sensing is compared with Softcast in the wireless multi-user channel conditions. Then the effect of video reconstruction in different Gaussian noise and packet loss rate is compared by experiment. The results show that very good results can be achieved in the transmission frame.