像素域Wyner-Ziv视频解码端的相关噪声建模研究.pdfVIP

第十五届全国图像图形学学术会议 像素域Wyner-Ziv 视频解码端的 相关噪声建模研究 尹明,蔡述庭,章云,谢云 (广东工业大学自动化学院，广州，510006) 摘 要 ： 针对现有Laplace 模型不能精确描述相关噪声分布，分布式视频编码(distributed video coding,DVC)的率失真 性能改善非常有限，本文深入研究了量化噪声因素对边信息的影响，在现有模型中考虑量化噪声因素，修正Laplace 模型参 数，提出一种新的像素域分布式视频编码系统的相关噪声模型。实验结果表明改进模型能更精确地描述相关噪声的统计特征， 基于该模型的DVC 率失真性能优于基于Laplace 模型的DISCOVER 方案，增益最大可达0.78dB。 关键词 ：无线传感器网络; 分布式视频编码; 相关噪声; 量化噪声; 率失真性能 Correlation Noise Model at the Decoder for Pixel Domain Wyner-Ziv video coding YIN Ming, CAI Shu-ting,ZHANG Yun, XIE Yun (Faculty of Automation, Guangdong Univ. of Tech., Guangzhou, 510006) Abstract ：Due to the inaccuracy Laplace model for describing the distribution of the correlation noise, distributed video coding(DVC) achieved the limit improvement in rate-distortion performance. The quantization noise is taken into account for modeling the virtual channel at the decoder of Wyner-Ziv video coding in this paper. The model parameter is fixed so that the model can accurately describe the statistics of correlation noise. The experimental results show that the proposed model can character the correlation noise more accurately and the performance of DVC based on proposed model outperforms the DISCOVER based on the Laplace one with up to 0.78 dB. Keywords: wireless sensor network; distributed video coding; correlation noise; quantization noise; rate-distortion performance [4,5] 制的应用需求，引起了人们强烈的兴趣 。DVC 是 1 引言 建立在分布式信源编码(DSC)理论之上的一种视频 编码方法，Wyner-Ziv 编码是其有损编码形式。通 随着以无线多媒体传感器网络(Wireless