基于多特征的帧级新型混合屏幕内容编码工具决策算法

Decision Algorithm of a New Frame-Level Hybrid Screen Content Encoding Tool Based on Multiple Features

针对我国音视频第三代标准(AVS3)参考软件中单特征的帧级屏幕内容编码工具决策算法导致新型混合屏幕内容图像编码效率降低的局限性,提出了一种基于哈希命中率和哈希突变率多特征的帧级屏幕内容编码工具自适应决策算法.实验结果表明,对于选定的新型混合屏幕内容图像测试数据集,与现有AVS3标准参考软件相比,提出的改进算法在编码复杂度几乎不变的情况下,对于有损全帧内配置,Y分量、U分量、V分量平均BD-rate降低率分别可达1.34%、1.54%、1.80%,充分验证了改进算法的有效性.

In the reference software of China’s audio-video third-generation standard(AVS3),the single-feature frame-level screen content coding tool decision algorithm leads to reduced coding efficiency for the new hybrid screen content images.Thus,based on the hash hit rate and hash mutation rate,the paper proposes a multiple-feature frame-level screen content coding tool decision algorithm.Compared with the existing AVS3 standard reference software,the experimental results show that for the selected novel hybrid screen content image test dataset,the average BD-rate reduction rates of the Y-component,U-component,and V-component can reach 1.34%,1.54%,and 1.80%,respectively,for all lossy full-frame configuration,with almost negligible coding complexity.