摘要
首先对JVT-G012码率控制算法进行了研究,针对原有算法中MAD线性预测的不准确性和对视频序列局部运动信息的不敏感性进行了算法改进。利用线性加权的块直方图差值(BHOD)作为视频序列的帧复杂度度量手段,用BHOD复杂度因子代替线性预测MAD作为复杂度手段,并将复杂度因子作为帧层目标比特数分配的加权系数。实验数据表明:与参考算法相比,改进算法对于不同运动剧烈程度的序列有良好的表现,能够更加精确地控制码率,峰值信噪比(PSNR)得到了明显提高,序列帧间质量波动减小,在解码端的视频序列稳定性较好。
Firstly, the JVT-GO12 rate control algorithm is studied, and the previous algorithm is improved aiming at the inaccurate linear predicted- MAD and the insensitive local sports information of video sequences. By using linear weighted block histogram difference （BHOD） as frame complexi- ty measure of video sequences, linear predicted MAD is substituted by BHOD complexity factor. The complexity factors are further used as the weigh- ted coefficient of frame target bit distribution. Experimental data show that ： Compared with the referenced algorithm, the improved algorithm has good performance for different sport intensity of sequences. The proposed algorithm can effectively control bit rate, and the PSNR has been obviously improved. Quality fluctuation between frames of sequences is reduced and the stability of decoded video sequences is quite better.
出处
《电视技术》
北大核心
2014年第19期8-12,共5页
Video Engineering
基金
国家自然科学基金项目(61171126)
关键词
帧级
线性预测模型
线性加权
BHOD
frame-level
linear prediction model
linear weighted
BHOD
