Study On Rate Control In H.264/AVC Video Coding Standard

Video Communications require very much bandwidth, comparing with other information types such as test and data. With the enormously growth of the requirement of Digital Video and Image transference , voice and Data Communication can not satisfy the people any more. In order to adapt the bandwidth-limited channels, especially wireless channels, video source must be compressed firstly.The early video coding standard have many drawbacks obviously due to the technology limitations. So they can not achieve high quality and low delay under narrow bandwidth. ITU-Tand ISO/IEC join to establish and release a new generation international video coding standard: H.264/AVC. It uses the Rate Distortion Model to control the bit rate , and to achieve the best efficiency of compression under the situation when the bit rate is restricted. In this thesis, the rate control algorithm is been investigated.Firstly, we introduce the H.264/AVC coding standard and analyse its rate control method.Secondly, we modify the rate control scheme of I-frame. By using a new way to estimate the target bitrate and distortion of I-frame, we can achieve better control of buffer state and the complixity of frames, optimize the calculation of the quantization parameter of each I-frame, and improve the visual quality.Thirdly, we propose PSNR as a new measure of P-frame to improve the estimation of complexity and resolve the problem that the mean absolute difference (MAD) of residuals is difficult to obtain a good estimate of complixity when the scene change or high motion in the H.264/AVC standard. Then we calculate the Lagrangian cost to balance the rate and distortion, by which we can find the best encoding mode.Finally, for the use of embed system, we port and optimize the whole H.264/AVC encoder to the ARM926EJ system. Also we test some video sequenceson it to verify the encode quality .The test result shows that we can achieve better control of the bits of I frame and the estimation the complixity and the target bits of the P frame. Then we can reduce the occupancy of buffer , avoid the overflow of buffer , encode the video sequence with less skipped frames and get better visual quality.