合成孔径雷达(SAR)在成像过程中由于回波的相干性而产生斑点噪声,严重干扰了图像的自动解译。根据SAR图像斑点噪声的特殊性,首先对影像进行对数变换,并使用àtrous运算法则对变换后的影像进行分解,然后通过贝叶斯原理估计修正小波系数。与自适应局域统计滤波和基于M allat分解的滤波算法进行分析比较,结果表明,该方法在噪声滤除和边缘保持方面效果较好。 Synthetic aperture radar (SAR) in the imaging process due to the coherence of the echoes generated speckle noise, seriously interfere with the automatic interpretation of the image. According to the particularity of the speckle noise of SAR images, the image is first transformed logarithmically and the transformed image is decomposed using àtrous algorithm, and then the wavelet coefficients are estimated by Bayesian principle. Compared with the adaptive local statistical filtering and the filtering algorithm based on allat decomposition, the results show that this method is better in noise filtering and edge preserving.