各向异性扩散的遥感图像边缘增强方法

为解决遥感图像边缘增强过程中辨识弱边缘和噪声的问题,提出一种改进的模糊各向异性边缘增强方法。根据非下采样轮廓波变换系数分布特征,获得像素几何结构信息;并基于各像素在不同子带的系数分布特征和噪声方差,分析其均值和最大值的模糊隶属度;利用模糊推理计算扩散系数,更好地控制各向异性扩散过程。实验结果显示,该方法具有更好的边缘增强和抑噪性能,能有效地辨识弱边缘和降低时间复杂度。

Edge enhancement using fuzzy anisotropic diffusion for remote sensing image

To solve the problem of distinguishing noise from weak edges, this paper proposed a novel fuzzy anisotropic diffusion approach for remote sensing image edge enhancement. At first, it gathered the geometrical information pixel by pixel from the NSCT coefficients. Then, it obtained the mean and max fuzzy membership values by analyzing the distribution of coefficients and noise variance in different sub-bands. At last, calculated the diffusion coefficients through fuzzy inference and embedded it into anisotropic diffusion to better control on the diffusion processing. Experiments show that the proposed method has better edge enhancement and de-noised performance and efficiently preserve the weak edges and reduce the time complexity.