Rate control algorithm for high efficiency video coding (HEVC)

Abstract

Video coding is the process of compressing and decompressing a digital video signal, which has been widely used in TV broadcasting, video streaming, video conferencing and DVD applications. Considering the output bit rate of the video encoder varies over time due to the changes of scene activities, rate control has been introduced to adjust the bit rate of the compressed video signal such that the storage memory size or bandwidth limitation can be met properly. Therefore, rate control is vital in ensuring the successful transmission of coded video data through narrow-band or time varying channels in visual communication. Reviewing the current rate control models for video coding standards, such as reference model (RM) for H.261, the adaptive quantization algorithm for MPEG-2, the test model nearterm (TMN) for H.263, the verification model(VM) for MPEG-4 and the joint model (JM) for H,264 etc., research on rate control for High Efficiency Video Coding (HEVC), an emerging video coding standard with significant improvement over the previous standards is still under studied. In this thesis the problems of the rate control schemes designing for HEVC are mainly addressed. Starting from the nature of video frame’s characteristics, we first present a theoretical analysis on intra coding using convex optimization and then we present an adaptive complexity measure based rate quantization model for intra frames considering the gradient values and the details ratio. For inter frames, after exhaustively video sequences encoding, a strong linear relationship between the pre-analysis SATD (PSATD) value and the generated Bpp (bit per pixel) is revealed. Thus, a rate control scheme for HEVC with parameters updating adaptively is proposed accordingly. Apart from that, application of gradient based Bpp (GBpp) as an indicator for the first Quantization Parameter (QP) selection is introduced in the proposed scheme. Furthermore a new source rate model and distortion model are established by using the PSATD and Quantization Step Size (QStep). By minimizing the cost function built based on the proposed rate-distortion model, the optimization problem is solved properly. According to the optimal solution, a bit allocation scheme with consideration of the PSATD and buffer status is formed. The experimental results illustrate that our proposed methods outperform the previous algorithms implemented in HEVC for both intra and inter frame rate control schemes. Besides rate control, this thesis also contains hardware oriented designs for alternative transform of HEVC based on the mathematical constraints and an adaptive bilateral filter considering local characteristics.

Date of Award	2012
Original language	English
Awarding Institution	The Hong Kong University of Science and Technology