一种基于形态变换的小波分解运动补偿编码新算法

基于图象小波变换系数的多分辨率运动估值／补偿算法在信噪比、压缩率和视觉质量方面都比传统的＂时域运动补偿＋ＤＣＴ”算法好，特别是对图象多分辨率和传输多码率的支持非常适合现今通信网的要求。但是算法中巨大的运动估值运算量制约了它的实际应用。使用潜在运动区进行运动估值是减少运算时间的可行方法之一。我们根据图象序列的运动内容在小波变换域呈现的方向性和局部性的特点，提出了一种基于形态变换的小波分解运动补偿新算法。该算法利用形态变换运算提取潜在运动区，使运动估值只集中在运动发生的区域进行，在保持一定的补偿误差时大大降低了运动估值时间，实验结果证明了新算法的有效性。

A New Wavelet Decomposition motion Compensation Algorithm Based on Morphological Segmentation

Video compression based on the wavelet representation and multiresolution motion compensation (MRMC)has a superior performance in terms of the PSNR as well as the subjective quality.But the intensive computational consuming is the bottleneck.it is a feasible way to make an estimation by using the potential motion area(PMA) to locate the regions where motion has likely occurred before the actual motion.Considering that the final results rely heavily onthe accuracy of PMA,a new PMA predicted a lgorithm based on morphological segmentation is proposed in this paper.The wavelet representation has the advantages of spati alorientation selectivity,localization in both space (or time) and frequency.Moreover the mathematical morphology efficiently deals with geometrical features.The hierarchical morphological segmentation can make a tradeoff between the quality of the MRMC and the computational cost.Experimental results show that the proposed algorithm can reduceprocessing time significantly.Using the energy decision of displacement frame difference can make the motion detection less sensitive to noise.The morphologicalsegmentation and marking PMA in all the wavelet subimages can avoid the error propagation.