Distributed Compressed Video Sensing with Joint Optimization of Dictionary Learning and l1-Analysis Based Reconstruction

Fang TIAN  Jie GUO  Bin SONG  Haixiao LIU  Hao QIN  

Publication
IEICE TRANSACTIONS on Information and Systems   Vol.E99-D   No.4   pp.1202-1211
Publication Date: 2016/04/01
Publicized: 2016/01/21
Online ISSN: 1745-1361
DOI: 10.1587/transinf.2015EDP7373
Type of Manuscript: PAPER
Category: Image Processing and Video Processing
Keyword: 
compressed video sensing,  realistic signal characteristics,  dictionary learning,  alternating direction method with multipliers,  

Full Text: PDF(1MB)>>
Buy this Article



Summary: 
Distributed compressed video sensing (DCVS), combining advantages of compressed sensing and distributed video coding, is developed as a novel and powerful system to get an encoder with low complexity. Nevertheless, it is still unclear how to explore the method to achieve an effective video recovery through utilizing realistic signal characteristics as much as possible. Based on this, we present a novel spatiotemporal dictionary learning (DL) based reconstruction method for DCVS, where both the DL model and the l1-analysis based recovery with correlation constraints are included in the minimization problem to achieve the joint optimization of sparse representation and signal reconstruction. Besides, an alternating direction method with multipliers (ADMM) based numerical algorithm is outlined for solving the underlying optimization problem. Simulation results demonstrate that the proposed method outperforms other methods, with 0.03-4.14 dB increases in PSNR and a 0.13-15.31 dB gain for non-key frames.