摘要: |
在头肩视频低码率编码的情况下,需要准确判定不编码宏块,从而节省码字。本文描述了一个基于运动矢量扫描的新算法,能够预测运动区域和静态背景区域,并在保证感兴趣区域质量的情况下,对静态区域的大部分宏块不编码。实验结果显示在低码率下,该算法能够大大节约编码的码字,而且在小运动序列时,可以显著减少编码的时间。 |
关键词: 视频编码 头肩视频序列 低码率编码 宏块 运动矢量扫描 |
DOI:10.3969/j.issn.1001-893X. |
投稿时间:2005-03-10 |
基金项目:电子科技大学校科研和教改项目 |
|
A New Method for Reducing Bit-Rate in Head and Shoulder Video Sequences |
|
() |
Abstract: |
In order to save bits,skipped macroblocks(MB) need to be determined accurately in head and shoulder video sequences encoded at low bitrates.This paper presents a new algorithm based on motion vector scanning that can predict motion region and static background region,and so it may skip many MBs in background on condition that region of interest may keep rate-distortion performance.Experimental results show that the algorithm can save more bits at low bit-rates and greatly reduce the encoding time when small motion video sequences are encoded. |
Key words: Video encoding,Head and shoulder video sequences,Low bit rate encoding,Macroblock,Motion vector scanning, |