论文部分内容阅读
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。
The multi-resolution motion estimation / compensation algorithm based on image wavelet transform coefficients is better than the traditional “time domain motion compensation + DCT” algorithm in terms of signal-to-noise ratio, compression rate and visual quality, Support for transmitting multi-rate is well suited to today’s communications network requirements. However, the huge amount of motion estimation in the algorithm restricts its practical application. Using potential areas of motion for motion estimation is one of the possible ways to reduce computation time. According to the directional and local characteristics of motion sequences of image sequence in the wavelet transform domain, a new algorithm of wavelet transform motion compensation based on morphological transformation is proposed. The algorithm uses the morphological transformation to extract the potential motion area, which makes the motion estimation only focus on the area where the motion occurs. It can greatly reduce the motion estimation time while maintaining a certain compensation error. The experimental results show the effectiveness of the new algorithm.