K-means算法在遥感分类中的应用

文中首先讨论了多光谱图像分类前需要进行的图像几何准和大气辐射校正等预处理方法，然后根据Ｋ－ｍｅａｎｓ算法的数理模型，重点讨论了分类的技术过程。最后讨论了从不同时相的图像变化中提取人为事件痕迹的方法，给出了ＴＭ图像的分类结果及事件判定结果。该方法在军事控卫星核查的普查阶段有较好的实际应用价值。

Application of K-means classification in remote sensing

The preprocessing method of geometrical image match and atmospherical image correction for multispectral image classification was briefly introduced, with the emphasis on image classification based on the mathematical and physical model of K-means. Finally, the method of picking-up man-made affairs from the variation of the image parameters in different time was discussed, and accordingly the TM image classification was attained and the affair was characterized. This method presented in this paper is useful in the investigation of nuclear test and arm control.