复小波域最大后验概率图像复原

MAP image restoration in dual-tree complex wavelet transform domain

提出了一种双树复小波变换域最大后验概率图像复原方法。该方法通过在最大后验概率图像迭代复原过程中构建噪声残差,并采用双树复小波变换零均值高斯模型对参差进行降噪处理,从而避免了原泊松最大后验概率图像复原过程中噪声放大的问题,实现了迭代复原的正则化目的。对比实验结果表明,该图像复原方法能很好解决恢复迭代中噪声放大的问题,同时,在视觉效果、PSNR、ISNR等指标上均比Wiener、Pisson-MAP等算法好。

A novel MAP image restoration method in Dual-tree Complex Wavelet Transfonn（DCWT） domain is proposed.The residual noisy sub-image in iterative process of Poisson-MAP restoration algorithm is set,and the noise in the residual noisy sub-image is suppressed by using of DCWT zero mean Gaussian distributing model denoising method.So,the problem of noise magnification in original Poisson-MAP restoration algorithm can be solved and the regularization purpose of iterative restoration algorithm is realized.Experimental results show that the proposed restoration algorithm is possible to achieve an excellent balance both restrains noise amplification effectively and preserves as many image details and edges as possible,especially under low SNR conditions.