基于纹理特征多分辨双Markov-GAR模型的SAR图像分割

为了提高SAR图像分割精度,提出在以灰度共生矩阵产生的纹理统计量为特征所生成的图像上,同时考虑SAR图像像素间空间分布特征和局部灰度均值和方差等统计量给出多分辨双Markov框架下的GAR模型,采用多分辨MPM的参数估计方法及对应的无监督分割算法,对SAR图像进行纹理分割。实验结果表明该方法用于一些高分辨SAR图像,与基于灰度图像上的多分辨双Markov-GAR模型纹理分割相比,在分割精度上能降低分割时的错分率。

Texture Feature-Based Segmentation of SAR Images Using a Multiresolution Pairwise Markov-GAR Model

This paper presents a new method for segmentation of synthetic aperture radar(SAR) images.We take into account spatial distributed characters between pixels of SAR images as well as the local means and variances statistics of gray level,a Gaussian autoregressive (GAR) model under a multiresolution pariwise Markov framework can be proposed based on texture feature images witch come from gray level co-occurrence probability statistics.For texture segmentation of SAR images,using the multiresolution maximization of the posterior marginals(MPM) estimate with the corresponding unsupervised segmentation algorithm on those texture feature images.Compared with multiresolution pariwise Markov-GAR model texture segmentation based on gray level images,for some SAR images,the result of experimentation showed that the method used in this paper has a better performance on segmentation precision.