概率模型下的K-means算法在SAR图像分类中的应用

Application of K-means Under Probability Model in Classification of SAR Image

遥感图像分类是将图像所有的像元按其性质分为若干个类的技术过程。常用的分类方法有监督分类和非监督分类。监督分类需要有足够的先验知识;非监督算法是按照某种相似性准则对样本进行合并或分类,所以并不需要有先验知识。但是传统的非监督分类算法存在着分类精度较低,分类结果比较粗糙等缺点。例如K-means算法、isodata算法等。提出一种基于高斯分布和瑞利分布两种概率模型的K-means算法,从结果可以看出,分类效果要明显优于传统的K-means算法。

The classification of remote sensing is a process that all the pels in the image are separated into some species by their characters. The normal ways are Surpervised classification and Un-supervised classification. Supervised classification need enough prior knowledge. Un-surpervised classification is a process of clustering, which is separated or incorporated by some rules, it does not require the prior knowledge.But the traditional un-surpervised classification has some falts,for example,the lower precision,the coarse result, such as K-means, isodata. In this paper, a method, K-means in the Gauss and Rayleigh Distribution is introduced. From the result, we can find that the impact is more clear than traditional K-means.