由于K-means 聚类方法对遥感图像进行分类时，对训练样本的选取依赖性很大，容易陷入局部最优的陷阱的情况，本文提出利用模拟退化算法对K-means 的聚类进行优化以获得

全局最优解的分类新方案。并以多波段影像为例进行验证分析，结果表明该方法可行，收敛

结果优于K-means 聚类算法，分类精度相对传统的K-means 算法更高。-Because K-means clustering classification depend on the training sample selection,great easy to fall into local optimum, in this paper , using Simulated Annealing algorithm to optimize K-means cluster analysis to obtain the global best optimal solution of the classification of 20 the new program. And multi-band image as an example to verify the analysis results, showing that the method is feasible, convergence results are better than K-means clustering algorithm, classification accuracy compared to traditional K-means algorithm is higher