外部非局部自相似先验的图像去噪

Image Denoising via External Non-local Self-similar Prior

针对在噪声水平比较高的情况下难以从噪声图像本身提取准确先验信息的问题,提出一种从外部干净图像数据集学习非局部自相似先验信息的图像去噪方法。首先用高斯混合模型学习外部干净图像的非局部自相似先验信息,其次利用最大后验概率估计的方法找到与噪声图像块最匹配的外部先验信息,最后利用外部先验对噪声图像块进行稀疏表示。实验对比表明,所提算法在去除噪声的同时可以较好地保留图像的细节信息,使图像数据集的平均峰值信噪比提高0.18 dB以上。

To solve the problem that it is difficult to extract accurate prior information from the noisy image itself when the noise level is relatively high,an image denoising method is proposed to learn non-local selfsimilar prior information from the external clean image dataset.Firstly,a Gaussian mixture model is used to learn the non-local self-similar prior information of the external clean image.Secondly,the maximum posterior probability estimation method is used to find the external prior information that best matches the noise image block.Finally,a sparse representation of noisy image blocks is achieved by external prior.Experimental results show that the proposed algorithm can better retain the details of the image while removing the noise,and improve the average peak signal-to-noise ratio of the image dataset by more than 0.18 d B.