中文摘要：

聚类分析是遥感图像非监督分类的有效方法，蚁群算法具有离散性和并行性的特点，蚂蚁觅食行为、蚂蚁堆积尸体行为和基于蚂蚁自我聚集行为的聚类算法是目前研究较为广泛的3种基于蚂蚁的仿生聚类算法．为验证上述3种算法的有效性，在对这3种聚类算法进行研究的基础上，针对遥感图像进行了聚类实验．实验结果表明，基于蚂蚁的聚类方法对图像的聚类分析是有效的，较传统的k均值和模糊C均值算法有一定优越性．

英文摘要：

Cluster analysis is the most efficient unsupervised classification method for remote sensing images. We studied three typical bio-inspired ant colony clustering algorithms because of their characteristics of discreteness and parallelization. They are based on the behavior of ants that are ： searching for food ; accumulating dead bodies ; self clustering behavior. In order to test the efficiency of these bio-inspired clustering algorithms, experiments were con- ducted using them on remote sensing images. The characteristics of these ant bio-inspired algorithms have been compared and analyzed, and the results of experiments indicate their adaptability and appropriate conditions for the use of these methods.