Compressed sensing mobile phone video communication based on structural similarity
|School||Taiyuan University of Science and Technology|
|Course||Pattern Recognition and Intelligent Systems|
|Keywords||Video transmission SSIM Distributed Video Coding Android|
Mobile phone video communication is a point-to-point videocommunication services, it can realize our dreams that call not only hear itssound but also see the person’s face which use the telephone network two-wayreal-time transmission the image and voice signals of both parties. Among them,the video signal compression coding, wireless transmission, decoding recoverytechnology is the key technology of video communication. In conventionalvideo coding, both MPEG and H.26x and other international compressionstandard are encoded in the motion estimation and motion compensation toremove the correlation between frames, therefore, the encoding process is morecomplex, video communication equipment is not suitable for low energyconsumption (such as mobile phone). Distributed video coding (DistributedVideo Coding, referred to as DVC) is the application of distributed sourcecoding in the video, a novel video coding framework is different from thetraditional video coding. First of all, because of the complexity of DVC transferencoding to the decoder, so in encode, the code become simple; secondly,because the DVC of a video frame using the independent coding framework, toovercome the traditional frame coding for joint coding caused by "the errorpropagation", robustness, have certain combat wireless transmission errorstherefore, especially suitable for mobile phone which video transmission, sensornetwork coding operation ability is not strong.This paper discusses the distributed mobile phone video communicationbased on compressed sensing, mainly completed the following work:First, study a novel technology that combining the compressed sensingwith DVC-Distributed compressed video sensing framework, using compressedsensing let it become further reduces the complexity of the DVC encoder withdata acquisition and compression.Second, this paper proposes an adaptive observation system of structuresimilarity based on compressed video sensing scheme. Because the video signal has the change between frames, for each video frame fixed observation ratecannot meet the optimization of video signal compression, therefore, tooptimize and observed the rate distribution of distributed compressed sensingusing the structure similarity, so that it can realize an adaptive observation ofcompressed sensing rate allocation, so as to improve the reconstructed videorate distortion performance.Third, this paper studies the overall framework of Android mobile phoneoperating system; in Android system, implementation of video communicationbased on H.264, and studied the use of distributed video communicationscompressed sensing video coding model, the preparation for the implementationof our proposed algorithm in the mobile phone on the future.