HEVC Quantization Parameter Selection Algorithm Based On The Inter-frame Dependency

Efficient video coding is a key basic technology for many digital media applications.The new generation HEVC(High Efficiency Video Coding)video coding standard uses multiple efficient coding technologies and shows excellent coding compression performance.With the same video coding quality,the compression efficiency is doubled compared to the previous generation video coding standard H.264.In actual applications,due to the limitations of transmission bandwidth and encoding complexity,the Rate Distortion Optimization(RDO)criterion is commonly used in video encoders to optimize the design of encoding algorithms.Therefore,researching the rate-distortion characteristics in HEVC is the key to fully tap the compression potential of the HEVC standard.Quantization is one of the most critical modules in a video encoder.Its task is to quantize the residual coefficients according to the selected quantization parameter(QP).On the one hand,the quantization parameters directly determine the absolute distortion and bit rate of the encoded video.On the other hand,the Lagrangian multiplier used in the rate-distortion optimization algorithm is also calculated from the quantization parameters and determines the distribution of coding distortion and bit rate by affecting the mode selection in the coding process.The frame-level quantization parameter is the basis for quantizing each block in video coding,and directly determines the distortion and bit rate of a frame to a large extent.Therefore,considering the new rate-distortion characteristics of HEVC,the frame-level quantization parameters selection algorithm needs to be improved for better encoding efficiency of HEVC encoder.Frame-level quantization parameters in video coding are determined by the rate control algorithm.In HEVC,a rate-distortion optimized rate control algorithm based on Lagrange multipliers is used to calculate the quantization parameters for each frame.The Lagrange multipliers used in HEVC are calculated from the target bit rate based on the assumption that each frame is independent of another.The encoding result shows that the calculated Lagrange multipliers obtained by independent rate-distortion model can only optimize the rate-distortion cost of current frame.However,due to the use of many new techniques in HEVC,especially more predictive coding techniques,the bit rate and distortion between different frames in HEVC are no longer statistically independent,such as: more inter-frame predictive coding modes,resulting in time-domain distortion propagation;high-order context entropy coding results in bit rate propagation between neighboring units in the spatial domain.The inherent characteristics of distortions and bit rate propagation effects make the assumption of statistical independency of frame-level rate and distortion in traditional rate control algorithms no longer valid,which makes the selection of frame-level quantization parameters that take the dependency of frame-level rate and distortion characteristics into account.It becomes the key to further improve the compression performance and optimize the global rate distortion of the HEVC encoder.To solve this problem,this paper explores the inter-frame rate-distortion-dependent characteristics of time domain,and studies the frame-level quantization parameter selection algorithm based on the inter-frame rate-distortion dependence.The research in this article focuses on solving these two problems: one is to analyze the generation mechanism of inter-frame dependencies in principle,and to quantitatively describe the inter-rate-distortion dependency;the other is to apply inter-frame dependencies to the calculation of the Lagrangian multiplier in the rate control algorithm.The main research work of this paper includes:1.Proposing a time-domain dependent rate-distortion model.By exploring the root cause of interframe dependency in HEVC and studying the relationship between coding distortion and reference frame reconstruction distortion and motion-compensated distortion,we build a timedomain distortion propagation model.Exploring the relationship between the quantization parameter change of current coding frame and its corresponding fluctuations of distortion and bit rate in the entire GOP according to the built RD fluctuation models.These two models provide support for dependent rate distortion optimized frame-level quantization parameter selection.2.Proposing a frame-level quantization parameter selection algorithm.By studying the global ratedistortion optimization algorithm based on inter-frame dependencies,we build the frame-level quantization parameters calculation algorithm that reflect inter-frame dependencies.Using the Lagrangian optimization algorithm,the established distortion fluctuation and bit rate fluctuation models are introduced into the rate distortion optimization algorithm,and the Lagrangian optimization multiplier based on the inter-frame dependency is obtained.Combining the relationship between quantization parameters and Lagrangian multipliers in HEVC,the quantization parameters based on inter-frame dependencies are calculated.The experimental results show that the frame-level quantization parameter selection algorithm proposed in this paper can effectively reduce BD-BR by 1.62% with the complexity basically unchanged.