用于H.264宏块预测模式决策的失真度量方法

H．264中常见的预测模式决策算法以简单但不大符合HVS（人类视觉系统）特性的SAD（残差绝对值和）或SSD（残差平方和）作为主要选择依据。首先了提出一种基于模糊积分的图像失真度量及其快速计算方法，该度量将评价图像失真作为1个多维评价过程。其次，使用该度量准则在H．264视频编码算法中进行宏块预测模式选择。实验结果表明，所提出的图像失真度量方法比较符合HVS特性，并且因为其计算简单，使得该度量能够用于诸如H．264等视频压缩领域中。

An Efficient Prediction Mode Decision Algorithm for H. 264 Video Coding Using Fuzzy Integral

The capability of the new H.264 video coding standard benefits much from using Intra and Inter predictions.However,the best prediction mode chosen by SAD(sum of absolute distortions) or SSD(sum of square distortions) may not be the really best one.In this paper,we aim to obtain the best prediction mode by proposing a new method which introduces a new image distortion metric based on fuzzy integral.In the full paper,we explain the new method in much detail;here we give only a briefing.First,we give the definition of the above-mentioned fuzzy metric by considering the partition character of macroblocks in H.264.Then the fast calculation algorithm of the proposed fuzzy metric is deduced.In addition,the prediction mode decision algorithm based on this metric is given.Based on experimental results from still image compression,we give the curve relating PSNR(peak signal noise ratio) to fuzzy metric;this curve shows that the fuzzy metric suits HVS(human visual system) fairly well;we believe this is the most important feature of our proposed new method.We also give experimental results for H.264 video compression with JM7.6(Joint Model);these results show that the proposed image distortion metric can reduce the bit rate without large change in coding time and PSNR.