摘要
文中提出了一种基于EM算法的相关噪声模型估计的Wyner-Ziv分级视频编码方法,基本层用复杂度相对较低的DISCOVER相关噪声模型参数估计方法,保证了基本的图像质量;增强层利用EM算法在线学习,建立相关噪声模型。实验结果表明,文中提出的噪声模型估计算法与DISCOVER的噪声模型相比,在相关噪声分布有拖尾现象时,能使编码器的率失真性能明显提高。
Wyner-Ziv scalable video coding method based on EM algorithm for correlation noise model estimation is presented in this paper.Low-complexity correlation noise model parameter estimation of DISCOVER is used in the base layer to guarantee the basic image quality.EM algorithm is used on online learning and setting correlation noise model.The results show that,compared with the DISCOVER's noise model,the rate distortion performance of the Wyner-ziv coder is improved with the noise model estimation algorithm proposed in the paper when the correlation noise has long tails.
出处
《南京邮电大学学报(自然科学版)》
2010年第6期54-58,共5页
Journal of Nanjing University of Posts and Telecommunications:Natural Science Edition
基金
江苏省高等学校研究生创新计划项目(CX07B_107z)
南京邮电大学青蓝计划(NY207077)资助项目