摘要
像素差分类（PDC）块匹配运动估值算法是一种考虑独立像素特征的运动估值算法，针对PDC算法的最优匹配不唯一的问题，本文提出了一种基于块运动最小相对位移约束的方法予以改进。本文还引入了双门限法以降低PDC算法的门限敏感性。实验表明本算法不仅克服了原算法的上述缺点，用于视频压缩编码时，还可使解码信噪比和主观质量有所改善。硬件实现的复杂性仅有很小的增加。
Pel Difference Classification (PDC) method for blockmatching motion estimaion is a method considering the behavior of individual pels in the search to find the best match. To avoid the existence of more than one point of optima match in PDC algorithm, the paper proposes a method based on the restrint of minimal relative displacement of blockmotion to get the best match. To reduce the threshold sensitivity of PDC blockmatching, a dual-threshold is introduced to improved the stability of the algorithm. The experiments show that the proposed algorithm not only resolves the problems in traditional PDC blockmatching, but also improves the decoding SNR and perceptual quality when it is used in video compression. The computation complexity and hardware requirement are only increased a little by this algorithm.
出处
《信号处理》
CSCD
1997年第2期155-160,共6页
Journal of Signal Processing
基金
国家自然科学基金
天津市青年科学基金
