一种基于Bayesian准则和MRF模型的SAR图像滤波方法

An Algorithm for SAR Image Filtering Based on Bayesian Principle and MRF Modeling

提出了一种新的斑点噪声的滤波方法,该方法基于小波分解的框架,采用小波域隐MRF模型,计算图像中表征像素纹理特征的隐状态参数的后验概率,以此后验概率作为权重因子对传统GammaMAP滤波算法进行改进。该算法被用于模拟SAR图像和真实ERS 1图像的Speckle抑制。试验结果表明,与传统Gam maMAP算法相比,本文提出的方法能更好地抑制Speckle,并保持图像细节。

Based on wavelet transform, a new Speckle reduction method is presented in this paper, and using wavelet-domain hidden Markov models to calculate posterior probability of state parameter, which describes the texture information of images. Then the posterior probability is used as filtering weighting factors in GammaMAP algorithm to reduce Speckle. The algorithm proposed was applied to both simulative images and ERS-1 SAR images. Compared with the original GammaMAP , the results indicate that this new approach can suppress Speckle and preserve more details as well.