首页 | 官方网站   微博 | 高级检索  
     

H.264到HEVC的低复杂度视频转码算法
引用本文:蒋炜,田翔,陈耀武.H.264到HEVC的低复杂度视频转码算法[J].浙江大学学报(自然科学版 ),2014,48(6):994-003.
作者姓名:蒋炜  田翔  陈耀武
作者单位:浙江大学 数字技术及仪器研究所,浙江省网络多媒体技术研究重点实验室,浙江 杭州 310027
基金项目:国家自然科学基金资助项目(40927001);浙江科学技术创新团队资助项目(2011R09021-02);中央高校基本科研业务费专项资金资助项目.
摘    要:针对H.264视频转码为下一代压缩标准高效电视编码(HEVC)视频过程中耗费大量时间的问题,提出一种低复杂度的视频转码算法.该算法充分利用H.264压缩视频流中包含的信息,基于图像复杂度和编码比特数之间的关系对每帧图像进行编码复杂度分析,并根据分析结果决定编码树单元的搜索深度范围;基于HEVC码流中Skip模式与H.264码流中各种模式的映射关系对Skip模式进行提前判决,通过对编码比特数的统计分析快速选择预测单元的对称与非对称分割模式;依据运动矢量的相似性,优化了HEVC运动估计过程中预测单元的搜索起点和搜索范围,进一步减少了转码过程的计算量.实验结果表明:该算法与参考算法相比转码速度获得了大幅提高,同时还保持了几乎相同的率失真表现.


Low complexity video transcoding algorithm from H.264 to HEVC
JIANG Wei,TIAN Xiang,CHEN Yao-wu.Low complexity video transcoding algorithm from H.264 to HEVC[J].Journal of Zhejiang University(Engineering Science),2014,48(6):994-003.
Authors:JIANG Wei  TIAN Xiang  CHEN Yao-wu
Abstract:A low complexity video transcoding algorithm from H.264 video to the next generation video compression standard the high efficiency video coding (HEVC) was proposed in order to reduce the computational time consumed during the transcoding process. By making full use of the H.264 compressed video information included in the bitstream, the coding complexity of each frame was analyzed based on the correlation between image complexities and coding bits, and then the searching depth range was adaptively decided based on the analysis results. By exploiting the mapping correlations of Skip mode in HEVC and all modes in H.264, an early decision of Skip mode was then introduced. Through the statistical analysis of coding bits, a fast prediction unit partition selection for both symmetric and asymmetric partitions was described. Finally, the motion estimation process was optimized according to the motion similarity between H.264 and HEVC bitstreams. Experimental results show that, compared to the reference methods, the proposed algorithm achieves a significant improvement on transcoding speed, while maintaining similar rate-distortion performance.
Keywords:
点击此处可从《浙江大学学报(自然科学版 )》浏览原始摘要信息
点击此处可从《浙江大学学报(自然科学版 )》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司    京ICP备09084417号-23

京公网安备 11010802026262号