小波变换压缩图像的贝叶斯迭代后处理算法

A Bayesian Iterative Postprocessing Algorithm for DWT-Based Compressed Image

提出一种Bayesian框架下的压缩图像后处理算法,去除压缩图像的模糊现象和“振铃”失真.与以往方法不同的是该算法考虑到了在高压缩比情况下的模糊效应,应用模糊过程加偏差来等效压缩过程,并提出了相应的后处理模型.采用交替迭代的方法进行求解,并讨论了参数的选取.实验结果表明,对于高压缩比小波变换的压缩图像,该算法在主观视觉效果和客观标准的评价上都取得了良好的效果.

In this paper, a new postprocessing method based on Bayesian theory is proposed to alleviate the blurring effect and remove the ringing effect of a compressed image. Different from traditional method, blurring effect is considered especially at low bit-rate. The compression is equivalent to the blurring processing followed by adding bias, based on which the new postprocessing model is established. An iterative solution is proposed after proper simplification and the parameter selection is discussed. At last, lots of numerical experiments are performed, and the results show that both the subjective perceived quality and the objective evaluation are improved.