摘要
对MEPG-4的FGSS编码进行了改进并用于H.264/AVC编码中,提出了一种modifiedFGSS(MFGSS)的编码方案。该方案首先使高、低分辨率下的视频质量都具有精细可伸缩性,并使不同分辨率下的运动矢量建立了联系,既降低了高分辨率码流中的运动矢量编码所用的比特数,又当高分辨率码流中的运动矢量受损时可以通过低分辨率的运动矢量得到恢复。实验表明:编码效率比FGSS编码有了明显的改善。在此基础上,对基于H.264/AVC的MFGSS编码方案在现有PC平台上进行了实时优化。先进行了基于简化RDO准则和多媒体指令集对编码器进行的优化,基于实时性,又进行了MFGSS的2种结构优化,提出了2种方案。方案1可使编码速度平均提高72.76%,高分辨率视频的码率平均提高23.3%,信噪比(PSNR)平均降低0.23dB;方案2使编码速度平均提高10.95%,高分辨率视频的码率平均提高2.7%,而PSNR基本不下降,方案2是方案1的一种折衷。
After the optimization of H. 264-based FGSS,we propose an approach named modified FGSS（MFGSS）. The H. 264-encoded video has the characteristics of fine-granular scalability at each resolution such as OlF or QOIF. According to the relationship between the motion vectors for low spatial resolution and that for high spatial resolution ,the bits to code motion vectors can be reduced because when the motion vectors with high resolution are lost, those with low resolution can be used to compensate them. Experiments show our approach has better performances on coding efficiency in comparison with MPEG-4 FGSS. Next,we describe the approaches to realize real-time coding on PC platform. After the simplified RDO optimization and multi-media instruction-set optimization, two schemes are proposed to enhance the structure of coder to realize real-time coding. The experiments show that in the case of the first scheme,the coding speed increases 72.76% than before and the bit rate also increases 23.3 % and the coding PSNR reduces only 0.23 dB. However, in the case of the second scheme,the coding speed increases 10.95 % and the bit rate also increases 2.7 % and the PSNR reduces few. So the second scheme is preferable.
出处
《光电子．激光》
EI
CAS
CSCD
北大核心
2006年第8期948-952,共5页
Journal of Optoelectronics·Laser
基金
国家自然科学基金重点资助项目(60332030)
上海市重点攻关资助项目(055115008)
上海市重点学科基金资助项目(T0102)
