基于模板匹配整帧屏幕视频帧内无损编码方法

当前主流的视频压缩标准,如H.264、HEVC等,主要是针对摄像机拍摄的视频而设计的.而屏幕内容(SC)的视频是由计算机生成的,无噪声,要求更高的压缩效率,同时与摄像机拍摄的视频相比包含更多的结构相似性.提出了一种利用屏幕视频中相似性的模板匹配帧内编码方法,采用由21个像素点组成的模板,以像素点为最小单位对整帧进行模板匹配,并使用哈希函数实现高效的模板查找,最后对残差图像进行熵编码.实验结果显示,与标准的HEVC扩展方法相比,该方法在无损编码上压缩率可以提高一倍,编码时间减小了近一半.

Lossless intra coding on whole frame for screen content video based on template matching

The-state-of-art video compression standards, such as H.264 and high efficiency video coding (HEVC) are mainly designed towards the camera-captured video. However, screen content (SC) video is created by the computer without noise, which needs higher efficiency than the camera-captured video. SC picture contains more structural similarity in contrast to camera-captured picture which could be used to improve the intra-picture compression ratio. A template matching method (TMM) was proposed by exploiting this similarity for SC intra lossless picture. The 21 pixel points template and the high efficiency 24-bits hash table were designed in TMM intra frame coding. For every picture, the first step was to use template matching for every pixel, and then speed up the matching processing by hash function, finally encode the residual picture by entropy coding. Experimental results show that TMM can double the lossless compression ratio compared to HEVC range extensions version 4.1 and reduce the compression time by half.