摘要
多视点视频编码(Muti-view Video Coding,MVC)标准采用预测关系固定的编码结构,不能根据视频序列调整预测关系,导致无法获得更好的编码效率.本文针对这一缺陷提出一种确定预测结构的方法,首先对图像进行下采样获得粗粒度图像,然后利用粗粒度图像间的灰度方差获得灰度相近的视点,最后对所选取的灰度相近的视点运用全局视差确定I-视点.经实验验证,本文的方法可以提高编码效率并减少编码中的缓存容量.
An improved Multi-view Video Coding prediction structure is proposed to adapt to different videos. Firstly, coarse-grained picture was got with down-sampling; then, the views with similar gray using Gray-scale Variance are fought; finally, I-view is chosen by calculating global parallax between each of the views with similar gray. Experiments results show that the proposed method can improve coding efficiency and decrease cache for coding.
出处
《微电子学与计算机》
CSCD
北大核心
2012年第5期62-66,共5页
Microelectronics & Computer
基金
国家自然科学基金项目(60970069)
西北工业大学研究生创业种子基金(Z2001134)
关键词
多视点视频编码
预测结构
灰度均值
灰度方差
Multi-view Video Coding
prediction structure
mean gray
gray-scale variance
