High Efficiency Video Coding Intra Prediction Optimization Algorithm Based on Region of Interest
-
摘要: 针对高性能视频编码(HEVC)帧内预测编码算法复杂度较高的问题,该文提出一种基于感兴趣区域的高性能视频编码帧内预测优化算法。首先,根据图像显著性划分当前帧的感兴趣区域(ROI)和非感兴趣区域(NROI);然后,对ROI基于空域相关性采用提出的快速编码单元(CU)划分算法决定当前编码单元的最终划分深度,跳过不必要的CU划分过程;最后,基于ROI采用提出的预测单元(PU)模式快速选择算法计算当前PU的能量和方向,根据能量和方向确定当前PU的预测模式,减少率失真代价的相关计算,达到降低编码复杂度和节省编码时间的目的。实验结果表明,在峰值信噪比(PSNR)损失仅为0.0390 dB的情况下,所提算法可以平均降低47.37%的编码时间。Abstract: For the high complexity of High Efficiency Video Coding (HEVC) intra prediction coding algorithm, an HEVC intra prediction optimization algorithm based on Region Of Interest (ROI) is proposed. Firstly, the algorithm divides the Region Of Interest and Non-Region Of Interest (NROI) of the current frame according to image saliency; Then, the final grading depth of the current coding unit is determined by the proposed fast Coding Unit (CU) partitioning algorithm based on spatial correlation in the ROI, and the unnecessary CU partitioning process is skipped. Finally, the proposed Prediction Unit (PU) mode fast selection algorithm is used to calculate the energy and direction of the current PU based on the ROI, and the current PU prediction mode is determined according to the energy and direction, and the correlation calculation of the rate distortion cost is reduced, Achieving the purposes of reducing coding complexity and saving coding time. The experimental results show that the proposed algorithm can reduce the coding time by 47.37% on average when the Peak Signal-to-Noise Ratio (PSNR) loss is only 0.0390 dB.
-
表 1 快速CU划分算法正确率和PU预测模式快速选择算法命中率(%)
序列 QP=22 QP=27 QP=32 QP=37 平均 Traffic 93.7/91.4 95.6/92.3 96.1/95.6 96.8/96.1 95.6/93.9 BQTerrace 93.1/89.7 94.8/91.4 95.8/93.5 96.4/94.7 95.0/92.3 Partyscene 92.4/90.2 94.7/93.1 95.6/93.9 96.2/94.8 94.7/93.0 Blowing Bubbles 91.1/88.6 93.4/90.3 94.7/92.5 95.8/93.7 93.8/91.3 Johnny 92.3/89.8 94.6/92.7 95.3/94.5 96.1/95.3 94.6/93.1 平均 92.5/89.9 94.6/91.9 95.5/94.0 96.3/94.9 94.7/92.7 分辨率 序列 BDBR(%) BDPSNR(dB) $T$(%) $2560 \times 1600$ Traffic 0.7054/0.6874/0.6013 –0.0406/–0.0396/–0.0327 42.19/43.62/46.89 PeopleOnStreet 1.2017/1.1047/0.7161 –0.0593/–0.0617/–0.0410 43.94/45.05/50.14 $1920 \times 1080$ Kimono 0.6725/0.6435/0.6314 –0.0351/–0.0309/–0.0293 42.76/43.93/47.93 Basketball Drive 1.3316/1.2704/1.0341 –0.0296/–0.0311/–0.0274 43.35/44.86/48.19 Cactus 1.2073/1.3160/0.9758 –0.0314/–0.0348/–0.0317 41.87/45.16/48.34 $832 \times 480$ BQMall 1.1986/1.1476/0.7692 –0.0724/–0.0769/–0.0405 40.01/42.93/45.54 Basketball Drill 1.3843/1.2543/0.6963 –0.0716/–0.0683/–0.0317 39.16/43.47/46.74 RaceHorsesC 1.2196/1.1702/0.7163 –0.0631/–0.0574/–0.0385 40.54/43.24/45.83 $416 \times 240$ Keiba 1.4055/1.1394/0.5631 –0.0965/–0.0846/–0.0417 41.96/43.56/46.14 BQSquare 1.3423/1.2761/0.6176 –0.0913/–0.0877/–0.0475 41.64/44.87/46.86 BasketballPass 1.4063/1.4322/0.7568 –0.0714/–0.0793/–0.0513 43.45/44.14/47.43 $1280 \times 720$ FourPeople 0.9704/0.9417/0.6975 –0.0542/–0.0523/–0.0372 42.64/43.17/47.39 Vidy01 0.6725/0.6524/0.7351 –0.0403/–0.0443/–0.0462 41.47/41.83/46.87 Vidyo3 1.0457/0.9125/0.8143 –0.0562/–0.0549/–0.0496 42.09/42.54/46.13 平均 1.1260/1.0677/0.7375 –0.0581/–0.0574/–0.0390 41.93/43.74/47.17 表 3 本文算法与文献[13]算法实验结果对比
Class 文献[13]算法 本文算法 BDBR(%) BDPSNR(dB) $T$(%) BDBR(%) BDPSNR(dB) $T$(%) ClassA 0.9236 –0.0742 44.19 0.6697 –0.0392 48.62 ClassB 1.1747 –0.0557 45.77 0.8926 –0.0327 48.74 ClassC 1.3532 –0.0823 41.89 0.7369 –0.0354 45.86 ClassD 1.3479 –0.1022 43.94 0.6461 –0.0473 46.69 ClassE 1.0754 –0.0837 43.76 0.7493 –0.0441 46.93 平均 1.1750 –0.0796 43.91 0.7389 –0.0397 47.37 -
王莉, 曹一凡, 杜高明, 等. 一种低延迟的3维高效视频编码中深度建模模式编码器[J]. 电子与信息学报, 2019, 41(7): 1625–1632. doi: 10.11999/JEIT180798WANG Li, CAO Yifan, DU Gaoming, et al. A Low-latency depth modelling mode-1 encoder in 3d-high efficiency video coding standard[J]. Journal of Electronics &Information Technology, 2019, 41(7): 1625–1632. doi: 10.11999/JEIT180798 TAI Kuanghan, HSIEH M Y, CHEN Meijuan, et al. A fast HEVC encoding method using depth information of collocated CUs and RD Cost characteristics of PU modes[J]. IEEE Transactions on Broadcasting, 2017, 63(4): 680–692. doi: 10.1109/TBC.2017.2722239 LI Yue, YANG Gaobo, ZHU Yapei, et al. Unimodal stopping model -based early SKIP mode decision for high -efficiency video coding[J]. IEEE Transactions on Multimedia, 2017, 19(7): 1431–1441. doi: 10.1109/TMM.2017.2669863 TAI Kuanghan, CHEN Meijuan, LIN Jieru, et al. Acceleration for HEVC encoder by bimodal segmentation of Rate-Distortion cost and accurate determination of early termination and early split[J]. IEEE Access, 2019, 7: 45259–45273. doi: 10.1109/ACCESS.2019.2900517 HUANG Chao, PENG Zongju, CHEN Fen, et al. Efficient CU and PU decision based on neural network and gray level co-occurrence matrix for intra prediction of screen content coding[J]. IEEE Access, 2018, 6: 46643–46655. doi: 10.1109/ACCESS.2018.2866081 CHEN Meijuan, WU Yude, YEH C H, et al. Efficient CU and PU decision based on motion information for interprediction of HEVC[J]. IEEE Transactions on Industrial Informatics, 2018, 14(11): 4735–4745. doi: 10.1109/TII.2018.2801852 LIU Xingang, LIU Yinbo, WANG Peicheng, et al. An adaptive mode decision algorithm based on video texture characteristics for HEVC intra prediction[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2017, 27(8): 1737–1748. doi: 10.1109/TCSVT.2016.2556278 WANG Xinzhi, FANG Yifan, LI Changdi, et al. Static gesture segmentation technique based on improved Sobel operator[J]. The Journal of Engineering, 2019, 2019(22): 8339–8342. doi: 10.1049/joe.2019.1075 GONG Shenjian, LI Guangqiang, ZHANG Yongju, et al. Application of static gesture segmentation based on an improved canny operator[J]. The Journal of Engineering, 2019, 2019(15): 543–546. doi: 10.1049/joe.2018.9377 余映, 吴青龙, 邵凯旋, 等. 超复数域小波变换的显著性检测[J]. 电子与信息学报, 2019, 41(9): 2231–2238. doi: 10.11999/JEIT180738YU Ying, WU Qinglong, SHAO Kaixuan, et al. Saliency detection of wavelet transform in hypercomplexdomain[J]. Journal of Electronics &Information Technology, 2019, 41(9): 2231–2238. doi: 10.11999/JEIT180738 ZHANG Tao, SUN Mingting, ZHAO Debin, et al. Fast intra-mode and CU size decision for HEVC[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2017, 27(8): 1714–1726. doi: 10.1109/TCSVT.2016.2556518 PAN Zhaoqing, LEI Jianjun, ZHANG Yun, et al. Fast motion estimation based on content property for low-complexity H.265/HEVC encoder[J]. IEEE Transactions on Broadcasting, 2016, 62(3): 675–684. doi: 10.1109/TBC.2016.2580920 GU Jiawen, TANG Minhao, WEN Jiangtao, et al. Adaptive intra candidate selection with early depth decision for fast intra prediction in HEVC[J]. IEEE Signal Processing Letters, 2018, 25(2): 159–163. doi: 10.1109/LSP.2017.2766766