视频图像数据压缩中的自适应量化技术

Adaptive Quantization Technique for Video Compression

为了使视频图像编码器产生的比特流和信道相匹配,本文提出了一种DCT系数自适应量化技术。此种技术,根据人眼视觉系统和DCT域的能量分布,得到最适合的量化矩阵,用码率缓存器状态控制量化精度、调节量化等级,从而得到稳定的高质量的重建图像。实验结果表明,图像编码器输出所需的比特率比传统方法少,解码重建图像主观和客观评价得到改善,且图像质量稳定。

In order to generate a bit stream from the video enceder that matches the capacity of the intended video channel, this paper presented a new quantization technique for the DCT coefficients. The dynamic adaptive quantizer was designed according to the human visual system （HVS） and DCT domain energy distribution to optimize the decoded image quality. This technique involves automatically deriving the most favorable quantization matrix from the HVS and DCT domain energy distribution of image sequences and scaling the resulting matrix furthermore by considering the buffer status. Subjective and objective tests indicated that the quality of reconstructed image is improved considerably and the required bit rate is much less than that of the conventional method. Particularly, the image quality can be kept almost consistent.