摘要
在彩色图像的三维离散余弦变换压缩体系中,交流系数游程编码后存在大量的长零游程,传统的JPEG基于(Run,Level)统计进行熵编码,压缩性能较差;通过对交流系数分布的研究,提出基于(Run Level,Level)的联合概率分布进行熵编码的改进方法,该方法码表简单,且可直接复用JPEG码表,有很大潜力应用于更高维的视频图像变换压缩领域。另外,根据非零系数的分布特点和编码效率分析,对密集区和疏散区域采用不同的编码方法。实验结果表明,在相同的PSNR下,较JEPG基线标准有11%的码流节省。
In the compression system of three dimensional discrete cosine transform, there are lots of long zero run-lengths of alternating current coefficients, which are encoded by run length coding, traditional entropy coding method in JPEG based on the statistics of （Run, Level） is not suitable. By path of the study of distribution of AC, proposes the improved entropy coding algorithm based on jointly probability of （Run Level, Level）, the size of code table is moderate, code table in JPEG is multiplexed, and there is great potential in the higher dimensional transform compression field of video image in further. According to the non-uniformity of non-zero coefficient and the analysis of coding efficiency, the scattered and clustered areas of non -zero coefficients are coded with different methods. The experiment results show that, when the peak signal and noise ratio is the same, there is 11% code rate savings compared with the Baseline of JPEG standard.
