Belief propagation‐based compressed video sensing for mobile Internet. (29th July 2015)

• Record Type: Journal Article
• Authors: Tian, Fang; Liu, Haixiao; Song, Bin; Qin, Hao; Ren, Guangliang
• Other Names: Yan Zheng guestEditor.; Niemi Valtteri guestEditor.; Yang Laurence T. guestEditor.
• Is Part Of: International journal of communication systems. Volume 29:Number 16(2016)
• Page Start: 2427
• Page End: 2438
• Publication Date: 2015-07-29
• Subjects: compressed sensing -- sparse reconstruction -- belief propagation -- video communication -- mobile Internet; Telecommunication systems -- Periodicals; 621.382
• DOI: 10.1002/dac.3023
• Languages: English
• ISSNs: 1074-5351

Abstract

Summary: The very high data rate applications such as streaming high‐definition video have attracted much attention of researchers in 5G systems. However, the design of 5G communication terminals faces several challenges, including how to reduce the computational burden of signal acquisition at the terminals and how to improve both the reconstruction quality and robust performance in wireless channels. How simple and fast online video acquisition can be implemented and how the original images can be recovered effectively by an offline algorithm for mobile Internet are taken into consideration in this paper. Specially, we firstly propose a Markov chain model to characterize variations of the video sequences' temporal correlation, wherein we utilize the measurement residue that reflects the intensity changes to estimate the transition matrix. Then, a compressed video sensing framework for mobile Internet is presented with the belief propagation algorithm and the proposed video signal model. Numerical results show that our proposal could efficiently utilize the temporal correlation in video and then obtain both the improved reconstruction quality and robust performance in wireless communication systems. Copyright © 2015 John Wiley & Sons, Ltd. Abstract : How simple and fast online video acquisition can be implemented and how the original images can be recovered for mobile Internet are considered in this paper. Specially, we propose a Markov chain model to characterize variations of the video sequences' temporal correlation, wherein we utilize the measurement residue that reflects the intensity changes to estimate the transition matrix. Then, a compressed video sensing framework for mobile Internet is presented with the belief propagation algorithm and the proposed video signal model.