北京邮电大学学报

  • EI核心期刊

北京邮电大学学报 ›› 2010, Vol. 33 ›› Issue (5): 126-130.doi: 10.13190/jbupt.201005.126.lihw

• 研究报告 • 上一篇    下一篇

自适应预测残差和量化矩阵的码率控制

李宏伟,吴成柯,宋锐,仵小波   

  1. 西安电子科技大学 综合业务网理论及关键技术国家重点实验室
  • 收稿日期:2009-11-04 修回日期:2010-01-12 出版日期:2010-08-28 发布日期:2010-06-15
  • 通讯作者: 李宏伟 E-mail:lhw3000@163.com
  • 基金资助:

    国家自然科学基金资助项目(No.60802076);高等学校创新引智计划资助项目(B08038);长江学者和创新团队发展计划资助项目(PCSITR);国家级.国家自然科学基金项目

Rate Control for Adaptive Prediction Error and Quantization Matrix

  • Received:2009-11-04 Revised:2010-01-12 Online:2010-08-28 Published:2010-06-15

摘要:

为了进一步提高H.264中码率控制的性能,提出了自适应预测残差和量化矩阵的码率控制. 首先根据宏块梯度方向权值和简化率失真模型,确定是否对宏块进行离散余弦变换(DCT);然后通过新的比特分配模型计算宏块的量化参数,并通过坡度加权确定宏块的坡度系数,最终根据得到的量化参数(QP)和坡度系数计算宏块的量化矩阵,并使用该量化矩阵对宏块进行编码. 仿真结果表明,与H.264基本单元层码率控制算法相比,本算法在严格控制编码时间和输出码率的情况下,使重建图像的峰值信噪比(PSNR)值提高021dB.

关键词: 码率控制, 预测残差, 量化矩阵, 量化参数

Abstract:

A rate control algorithm for adaptive prediction error and quantization matrix is proposed in order to improve the performance of rate control in H.264 further. Firstly, the new algorithm determines whether discrete cosine transform (DCT) is applied to the macroblock according to the gradient directional weight of macroblock and to the simplified rate distortion model. Secondly, the quantization parameter (QP) is calculated through a new dual target bits allocation model, and then the slope coefficient is calculated by the weighted average of slope in the macroblock. Finally, the macroblock is quantized by the quantization matrix determined according to QP and slope coefficient. Simulation shows that, compared with rate control algorithm of basicunit in H.264, the proposed algorithm restricts coding time dramatically and controls bitrate precisely. Moreover, the peak signal noise ratio (PSNR) is increased with 021dB in the proposed algorithm.

Key words: rate control, prediction error, quantization matrix, quantization parameter