基于空-时域视觉敏感度的码率分配算法
|
摘要
新一代高效视频编码HEVC(High Efficiency Video Coding)标准中,λ域的码率控制采用自适应码率分配策略,其主要根据前一参考帧对应参考单元的码控参数来调整当前编码单元的码率分配参数。但是,这种码率分配策略下的编码结果不适合于人类视觉特性。提出一种基于空-时域视觉敏感度的码率分配策略,采用一种恰可感知失真JND(Just Noticeable Difference)模型来获取一帧图像空域上的最大视觉失真值。通过对最大失真值计算,得到每个编码树单元CTU(Coding Tree Unit)空域敏感度权重。利用相邻帧在时域上的变化,获取帧级的码率分配比例。根据帧间变化得到自下而上的视觉关注度,从而计算出每个编码树单元在时域上的敏感度权重。结合空域与时域的视觉敏感度,获得基于空-时域的码率分配权重,将码率按照视觉特性合理的分配到每个编码树单元。实验结果显示,该算法可以有效提高码控的主观质量,同时降低码率波动。
In the new generation of HEVC( High Efficiency Video Coding) standard,the adaptive bit rate allocation strategy is adopted for the rate control in λ domain. The bit rate allocation parameters of the current coding unit are adjusted according to the rate control parameters of the reference unit corresponding to the previous reference frame.However,the coding result under this bit rate allocation strategy is not suitable for human visual characteristics. In this paper,a bit rate allocation strategy based on spatial-temporal domain visual sensitivity was proposed. A JND( Just Noticeable Difference) model was adopted to obtain the maximum visual distortion in the spatial domain of a frame image. By calculating the maximum distortion value,the spatial sensitivity weight of each CTU( coding tree unit) was obtained. By using the change of adjacent frames in temporal domain,we got the frame level bit rate allocation ratio.The bottom-up visual attention was obtained according to the frame variation,and the sensitivity weight of each CTU in the temporal domain was calculated. Combining the spatial and temporal visual sensitivity,the weight of bit rate allocation based on spatial-temporal domain was obtained,and the bit rate was allocated to each CTU according to the visual characteristics. The experimental results show that the proposed algorithm can effectively improve the subjective quality of code control and reduce the rate fluctuation.
In the new generation of HEVC( High Efficiency Video Coding) standard,the adaptive bit rate allocation strategy is adopted for the rate control in λ domain. The bit rate allocation parameters of the current coding unit are adjusted according to the rate control parameters of the reference unit corresponding to the previous reference frame.However,the coding result under this bit rate allocation strategy is not suitable for human visual characteristics. In this paper,a bit rate allocation strategy based on spatial-temporal domain visual sensitivity was proposed. A JND( Just Noticeable Difference) model was adopted to obtain the maximum visual distortion in the spatial domain of a frame image. By calculating the maximum distortion value,the spatial sensitivity weight of each CTU( coding tree unit) was obtained. By using the change of adjacent frames in temporal domain,we got the frame level bit rate allocation ratio.The bottom-up visual attention was obtained according to the frame variation,and the sensitivity weight of each CTU in the temporal domain was calculated. Combining the spatial and temporal visual sensitivity,the weight of bit rate allocation based on spatial-temporal domain was obtained,and the bit rate was allocated to each CTU according to the visual characteristics. The experimental results show that the proposed algorithm can effectively improve the subjective quality of code control and reduce the rate fluctuation.
引文
[1] Doulamis N D,Doulamis A D,Konstantoulakis G E,et al.Efficient modeling of VBR MPEG-1 coded video sources[J].IEEE Transactions on Circuits and Systems for Video Technology,2000,10(1):93-112.
[2] Burnett I S,Pereira F. Digital Video:An Introduction to MPEG-2[M]. New York:Springer,1997.
[3] Vetro A,Sun H,Wang Y. MPEG-4 rate control for multiple video objects[J]. IEEE Transactions on Circuits and Systems for Video Technology,1999,9(1):186-199.
[4] Corbera J,Lei S. Rate control for low-delay video communications[S]. ITU Study Group,1997.
[5] Choi H,Yoo J,Nam J,et al. Pixel-wise unified ratequantization model for multi-level rate control[J]. IEEE Journal of Selected Topics in Signal Processing,2013,7(6):1112-1123.
[6] Li B,Li H Q,Li L,et al. Rate control by R-lambda model for HEVC[C]//Presented at 11th Meeting on JCTVCK0103,JCTVC of ISO/IEC and ITU-T,Shanghai,China,2012.
[7] Li B,Li H Q,Li L,et al.λdomain rate control algorithm for high efficiency video coding[J]. IEEE Trans. Image Processing,2014,23(9):3841-3854.
[8]陈伟国,陈芬,金德富,等.基于复杂度分析的HEVC帧内码率控制算法[J].计算机应用与软件,2017,34(1):143-148,174.
[9]荣倩倩,杨静.基于HEVC的LCU层码率控制算法改进[J].计算机应用与软件,2016,33(5):136-139.
[10]周建政.一种HEVC码率控制改进算法[J].计算机应用与软件,2016,33(8):183-184,259.
[11] Gao W,Kwong S,Yuan H,et al. DCT coefficient distribution modeling and quality dependency analysis based framelevel bit allocation for HEVC[J]. IEEE Trans. Circuits Syst. Video Techn.,2016,26(1):139-153.
[12] Li S,Xu M,Wang Z. A novel method on optimal bit allocation at LCU level for rate control in HEVC[C]//2015 IEEE International Conference on Multimedia and Expo(ICME).IEEE,2015.
[13] Li L,Li B,Liu D,et al.λ-domain rate control algorithm for HEVC scalable extension[J]. IEEE Transactions on Multimedia,2016,18(10):2023-2039.
[14] Lee J S,Ebrahimi T. Perceptual Video Compression:A Survey[J]. IEEE Journal of Selected Topics in Signal Processing,2012,6(6):684-697.
[15] Xu M,Deng X,Li S,et al. Region-of-interest based conversational HEVC coding with hierarchical perception model of face[J]. IEEE Journal of Selected Topics in Signal Processing,2014,8(3):475-489.
[16] Li S,Xu M,Deng X,et al. Weight-based R-λrate control for perceptual HEVC coding on conversational videos[J].Signal Processing:Image Communication, 2015, 38:127-140.
[17] Tang C W. Spatiotemporal Visual Considerations for Video Coding[J]. IEEE Transactions on Multimedia,2007,9(2):231-238.
[18] Zhao H,Xie W,Zhang Y,et al. An SSIM-motivated LCUlevel rate control algorithm for HEVC[C]//Picture Coding Symposium. IEEE,2014.
[19] Kim J,Bae S H,Kim M. An HEVC-Compliant Perceptual Video Coding Scheme Based on JND Models for Variable Block-Sized Transform Kernels[J]. IEEE Transactions on Circuits&Systems for Video Technology,2015,25(11):1786-1800.
[20] Chou C H,Li Y C. A perceptually tuned subband image coder based on the measure of just-noticeable-distortion profile[J]. IEEE Transactions on Circuits and Systems for Video Technology,2002,5(6):467-476.
[21] Li B,Li H,Li L. Adaptive bit allocation for R-lambda model rate control in HM[C]//JCT-VC of ITU-T SG16 WP3 and ISO/IEC JTC1/SC29/WG11 13th Meeting,Incheon,Rep.of Korea,Doc JCTVC-M0036,2013.
[22] Zhou Wang,A. C. Bovik,H. R. Sheikh,et al. Image quality assessment:from error visibility to structural similarity[J]. IEEE Trans Image Process,2004,13(4):600-612.
[2] Burnett I S,Pereira F. Digital Video:An Introduction to MPEG-2[M]. New York:Springer,1997.
[3] Vetro A,Sun H,Wang Y. MPEG-4 rate control for multiple video objects[J]. IEEE Transactions on Circuits and Systems for Video Technology,1999,9(1):186-199.
[4] Corbera J,Lei S. Rate control for low-delay video communications[S]. ITU Study Group,1997.
[5] Choi H,Yoo J,Nam J,et al. Pixel-wise unified ratequantization model for multi-level rate control[J]. IEEE Journal of Selected Topics in Signal Processing,2013,7(6):1112-1123.
[6] Li B,Li H Q,Li L,et al. Rate control by R-lambda model for HEVC[C]//Presented at 11th Meeting on JCTVCK0103,JCTVC of ISO/IEC and ITU-T,Shanghai,China,2012.
[7] Li B,Li H Q,Li L,et al.λdomain rate control algorithm for high efficiency video coding[J]. IEEE Trans. Image Processing,2014,23(9):3841-3854.
[8]陈伟国,陈芬,金德富,等.基于复杂度分析的HEVC帧内码率控制算法[J].计算机应用与软件,2017,34(1):143-148,174.
[9]荣倩倩,杨静.基于HEVC的LCU层码率控制算法改进[J].计算机应用与软件,2016,33(5):136-139.
[10]周建政.一种HEVC码率控制改进算法[J].计算机应用与软件,2016,33(8):183-184,259.
[11] Gao W,Kwong S,Yuan H,et al. DCT coefficient distribution modeling and quality dependency analysis based framelevel bit allocation for HEVC[J]. IEEE Trans. Circuits Syst. Video Techn.,2016,26(1):139-153.
[12] Li S,Xu M,Wang Z. A novel method on optimal bit allocation at LCU level for rate control in HEVC[C]//2015 IEEE International Conference on Multimedia and Expo(ICME).IEEE,2015.
[13] Li L,Li B,Liu D,et al.λ-domain rate control algorithm for HEVC scalable extension[J]. IEEE Transactions on Multimedia,2016,18(10):2023-2039.
[14] Lee J S,Ebrahimi T. Perceptual Video Compression:A Survey[J]. IEEE Journal of Selected Topics in Signal Processing,2012,6(6):684-697.
[15] Xu M,Deng X,Li S,et al. Region-of-interest based conversational HEVC coding with hierarchical perception model of face[J]. IEEE Journal of Selected Topics in Signal Processing,2014,8(3):475-489.
[16] Li S,Xu M,Deng X,et al. Weight-based R-λrate control for perceptual HEVC coding on conversational videos[J].Signal Processing:Image Communication, 2015, 38:127-140.
[17] Tang C W. Spatiotemporal Visual Considerations for Video Coding[J]. IEEE Transactions on Multimedia,2007,9(2):231-238.
[18] Zhao H,Xie W,Zhang Y,et al. An SSIM-motivated LCUlevel rate control algorithm for HEVC[C]//Picture Coding Symposium. IEEE,2014.
[19] Kim J,Bae S H,Kim M. An HEVC-Compliant Perceptual Video Coding Scheme Based on JND Models for Variable Block-Sized Transform Kernels[J]. IEEE Transactions on Circuits&Systems for Video Technology,2015,25(11):1786-1800.
[20] Chou C H,Li Y C. A perceptually tuned subband image coder based on the measure of just-noticeable-distortion profile[J]. IEEE Transactions on Circuits and Systems for Video Technology,2002,5(6):467-476.
[21] Li B,Li H,Li L. Adaptive bit allocation for R-lambda model rate control in HM[C]//JCT-VC of ITU-T SG16 WP3 and ISO/IEC JTC1/SC29/WG11 13th Meeting,Incheon,Rep.of Korea,Doc JCTVC-M0036,2013.
[22] Zhou Wang,A. C. Bovik,H. R. Sheikh,et al. Image quality assessment:from error visibility to structural similarity[J]. IEEE Trans Image Process,2004,13(4):600-612.