利用前景和背景分析提升H.264之壓縮效率

Abstract

本論文針對現行的視訊壓縮標準H.264，探討前景和背景分析的應用，目的在透過前景的影像品質和主觀視覺接受度提升、背景的資料量之降低，以及壓縮速度之提升等三方面，來改善整體標準的壓縮效率。由於H.264提供的編碼工具使用彈性很大，所以本論文在前景和背景區域已知的前提下，對於標準中可變區塊大小和失真模糊率的兩種編碼工具進行調整與分析，並提出兩個增加編碼效能的方法。同時，這些以前景和背景分析的編碼方法亦均能符合H.264標準內所規範的語法和語意。而實驗的結果也顯示，我們所提出的方法能相當有效的改善H.264的編碼效率。

The objective of this thesis research is to study possible improvements of the encoding performance of the H.264 video standard given the foreground and background models of the source video sequence. The encoding performance is measured in three aspects: the PSNR, bit-rates and coding speed. By analyzing the variable block-size and rate-distortion coding tools of H.264, we propose two different encoding methods for the video foregrounds and backgrounds to enhance the video quality of the foreground objects and to reduce the compression bit-rates of the background area while ensure that the syntax and semantics of the encoded bit-streams agree with H.264. Experimental results show the proposed methods are quite effective in the performance improvement of the H.264 encoding.