论文部分内容阅读
由于计算机生成的屏幕视频和摄像机拍摄的自然视频在本质上有很大差异,计算机屏幕视频具有特殊的图像特征,使用传统视频压缩工具对其进行压缩时,编码效果很不理想.提出一种算法,对屏幕视频中出现的近似像素进行合并处理,有效结合了基于串拷贝机制的字典编码方案,并应用于HEVC视频编码标准平台.近似像素合并的过程中,综合考虑了像素出现频度和像素差异等因素,在不损伤图像主观质量的前提下,该算法能明显改善屏幕视频编码方案的性能.实验结果表明,在全帧内(All Intra,AI)和低延迟(Low-delay B,LB)两种配置下,和ISC(Intra String Copy)算法比较,所提算法的码率平均节省了1.4%和0.35%.和HEVC屏幕视频编码测试模型SCM4.0比较,码率平均节省了10.39%和5.5%.
Because computer-generated screen video and nature video shot by the camera are different in nature, the computer screen video has special image features, and the traditional video compression tool is used to compress it, and the coding effect is not ideal.An algorithm , Combines approximate pixels appearing on the screen video and effectively combines the dictionary encoding scheme based on the string copy mechanism and applies to the HEVC video encoding standard platform.In the process of approximate pixel combination, the pixel appearance frequency and pixel The algorithm can significantly improve the performance of the screen video coding scheme without damaging the subjective quality of the image.The experimental results show that in all Intra (AI) and low-delay B ) In both configurations, compared with the ISC (Intra String Copy) algorithm, the proposed algorithm saves an average of 1.4% and 0.35% code rate.Compared with the HEVC screen video coding test model SCM4.0, the code rate is reduced by 10.39% And 5.5%.