caa a22 4500 60620539X CHVBK 20210128101007.0 cr unu---uuuuu 210128e20150701xx s 000 0 eng 10.1007/s00034-014-9951-x doi (NATIONALLICENCE)springer-10.1007/s00034-014-9951-x Band-Level Correlation Noise Modeling for Wyner-Ziv Video Coding with Gaussian Mixture Models [Elektronische Daten] [Ming Yin, Junbin Gao, Daming Shi, Shuting Cai] As one of the most adopted distributed video coding approaches in the literature, Wyner-Ziv (WZ) video coding is not yet on par with the motion-compensated predictive coding solutions with respect to rate-distortion (RD) performance. One of the essential reasons lies in the absence of reliable knowledge of the correlation statistics between source and side information. Most of the existing works assume a probability distribution of the statistical dependency to be Laplacian, which is not accurate but computationally cheap. In this paper, a correlation estimation based on Gaussian mixture model is proposed for the band-level correlation noise of discrete cosine transform domain Wyner-Ziv codec. The statistics of the correlation noise between WZ frame and corresponding side information is analyzed by considering the temporal correlation and quantization distortion. Accordingly, the model parameters for correlation noise are estimated offline and utilized online in consequent decoding. The simulation results of Kullback-Leibler divergence show that the proposed model has higher accuracy than the Laplacian one. Experimental results demonstrate that the WZ codec incorporated with the proposed model can achieve very competitive RD performance, especially for the sequence with high motion contents and large group of picture (GOP) size. Springer Science+Business Media New York, 2014 Correlation noise nationallicence Quantization nationallicence Wyner-Ziv coding nationallicence Rate-distortion nationallicence Gaussian mixture model nationallicence Wireless sensor network nationallicence Yin Ming School of Automation, Guangdong University of Technology, 510006, Guangzhou, China aut Gao Junbin School of Computing and Mathematics, Charles Sturt University, 2795, Bathurst, NSW, Australia aut Shi Daming School of Engineering and Information Sciences, Middlesex University, NW4 4BT, London, UK aut Cai Shuting School of Automation, Guangdong University of Technology, 510006, Guangzhou, China aut Circuits, Systems, and Signal Processing Springer US; http://www.springer-ny.com 34/7(2015-07-01), 2237-2254 0278-081X 34:7<2237 2015 34 34 https://doi.org/10.1007/s00034-014-9951-x text/html Onlinezugriff via DOI BK010053 XK010053 XK010000 Metadata rights reserved Springer special CC-BY-NC licence nationallicence 1 research-article jats NATIONALLICENCE NATIONALLICENCE NL-springer NATIONALLICENCE 856 40 https://doi.org/10.1007/s00034-014-9951-x text/html Onlinezugriff via DOI NATIONALLICENCE 700 1- Yin Ming School of Automation, Guangdong University of Technology, 510006, Guangzhou, China aut NATIONALLICENCE 700 1- Gao Junbin School of Computing and Mathematics, Charles Sturt University, 2795, Bathurst, NSW, Australia aut NATIONALLICENCE 700 1- Shi Daming School of Engineering and Information Sciences, Middlesex University, NW4 4BT, London, UK aut NATIONALLICENCE 700 1- Cai Shuting School of Automation, Guangdong University of Technology, 510006, Guangzhou, China aut NATIONALLICENCE 773 0- Circuits, Systems, and Signal Processing Springer US; http://www.springer-ny.com 34/7(2015-07-01), 2237-2254 0278-081X 34:7<2237 2015 34 34