摘要
形状编码是MPEG-4不同于其它标准(如MPEG—1和MPEG-2等)的新特点。本文提出改进的四叉树多分辨率的形状编码的算法,文中详细探讨了对原基于四叉树多分辨率形状编码算法改进部分:帧内边界宏块采用基于相关系数准则的四叉树编码,帧间边界宏块采用有效搜索区域的运动估计和运动补偿的四叉树编码。该算法用相关系数来控制四叉树的复杂度,客观视觉质量下降很少,而改进的四叉树的节点数明显下降,进一步提高形状编码效率。二值形状的帧间编码时,运动估计仅在有效搜索区域进行,减少搜索次数,极大提高搜索效率。处理结果表明与原四叉树的形状编码相比, 性能上更具竞争力。
Shape coding is new feature in MPEG-4 different from other standards, such as MPEG-1 and MPEG -2. This paper proposes a modified quad-tree mutli-resolution shape coding algorithm, and discusses in details the improved parts on original shape coding based quad-tree mutli-resolution, which include that quad-tree coding based in homogenous parameter rule is used for intra-frame boundary macroblocks, and motion estimation by searching only in the efficient search areas is used for inter-frame boundary macroblocks. In the proposed algorithm, the complex of quad-tree is controlled by the homogenous parameter, so that the objective quality lowers very little, the numbers of the modified quad-tree lowers greatly, and the efficiency of shape coding is improved more. The motion is estimated by searching only in the efficient search areas during the inter-frame shape coding of VOP, so as to cut down the numbers of searching and improve greatly the efficiency of searching.
出处
《信号处理》
CSCD
北大核心
2006年第3期361-365,共5页
Journal of Signal Processing
