基于感兴趣区域的HEVC压缩性能优化

ROI-based HEVC Video Compression Optimization

根据人类视觉系统(HVS)对纹理复杂及运动区域具有较强感知度的特点,提出了一种基于感兴趣区域的高效率视频编码(HEVC)压缩性能优化算法。首先使用Sobel梯度检测算子和运动矢量分别检测纹理复杂区域和运动区域,把检测到的纹理复杂及运动区域定义为感兴趣区域;再对感兴趣区域分级,通过调整量化参数(QP),优化比特分配。实验结果表明,与HEVC标准算法相比,所提算法码率平均减少了15.29%,时间平均节省了11.38%。

According to the feature that Human Visual System( HVS) has acute recognition of motion and complex texture,this paper presents a Region of Interest( ROI)- based High Efficiency Video Coding( HEVC) video compression algorithm. First,Sobel operator and motion vectors are applied to detect the complex texture area and motion area respectively which are,as a result,defined as ROI. Second,all the ROIs are graded,and then digital bits allocation is optimized by regulating the quantization parameter( QP). Experimental result shows that compared with the standard HEVC algorithm,the proposed algorithm brings down the coding rate by 15. 29% and working time by 11. 38%.