面向统计编码的联合比特分配算法

在统计编码系统中,需根据图像复杂度对各路节目进行联合比特分配,比特分配的准确性直接影响了图像质量.因此对图像复杂度的准确评估是统计编码的难点.鉴于传统的基于预测的算法对复杂度评估存在预测误差,提出了将原始帧作为参考帧,通过并行整像素运动估计计算各编码帧的SAD作为统计所需的复杂度信息,提高了图像复杂度评估的准确性,进而提高统计编码性能.通过测试CBR模式下和提出的统计算法下的视频码率以及图像质量来说明所提出算法的可行性.

Research of statistical coding algorithm based on parallel computing

In statistical coding systems,it need to distribute bits to each programs according to its image complexity. The accuracy of the bit allocation directly affects the image quality.Therefore, an accurate assessment of the image complexity is the biggest difficulty. Because the traditional algorithm based on prediction is an inaccurate way to evaluate image complexity, The paper propose a statistical coding algorithm,in which it calculate the SAD by parallel whole-pixel motion estimation using the original frame as reference frame,and treat SAD as the statistical information. In this way, it improve the accuracy of image complexity assessment and statistical coding performance.Experimental results demonstrate the superiority of the scheme by testing video bit rate and PQR under condition of CBR and VBR.