Font Size: a A A

Reasearch On Rate Control Based On Face Region

Posted on:2011-10-31Degree:MasterType:Thesis
Country:ChinaCandidate:L ZhuFull Text:PDF
GTID:2178360305460713Subject:Computer system architecture
Abstract/Summary:PDF Full Text Request
H.264/AVC video coding standard which is the newest video coding standard is paid close attention with its higher compression efficiency, better image quantity and network compatibility.Rate Control as a very important part of the video compressing encode,can compute QP and has responsibility to allcocate bit at the frame-level and macroblock-level.The paper mainly researches on Rate control algorithm of H.264/AVC at the low bandwidth include the following two areas:First, the video quality can not meet the need at the low Bandwidth, so researchers make some methods to increase the quality of ROI and hope to meet people's need using this method.A rate control algorithm based Region of interest is proposed in this paper, the algorithm which establishs an effective method in the ROI extraction, works at frame level and the basic unit level. The algorithm improves the tactics to allocate the bit at the frame level and imporves the MAD prediction at the basic-unit level. The algorithm also enhances the bit in the region of interest to raise the quality.the results shows the algorithm increase PSNR and improves the subjective quality.A rate control algorithm is proposed based on analyzing the lack of MAD prediction linear model.the algorithm predicts MAD using the imformation of time and space,and this way is more preciser than the original way which only use the imformation of space. The paper improves the tactics of bit-allocation at frame-level based on the new MAD prediction model.the new way of bit-allcocation can adaptively allocate the bit combining the buffer fullness and the coding complexity.the result shows the algorithm can increase PSNR.
Keywords/Search Tags:H.264, Rate Control, Region of Interest, MAD prediction, Bit allcacion
PDF Full Text Request
Related items