自适应蚁群和模糊聚类的SAR图像分割  

Segmentation of SAR images based on adaptive ant colony algorithm and fuzzy C-means

作　　者：鲍义东[1,2] 周改云[1] 赵伟艇[1] 

机构地区：[1]平顶山学院软件学院,河南平顶山467000 [2]哈尔滨工业大学机器人技术与系统国家重点实验室,哈尔滨150080

出　　处：《测绘科学》2016年第8期121-124,120,共5页Science of Surveying and Mapping

基　　金：河南省教育厅科学技术研究重点项目(14B520039);校青年科研基金项目(PXY-QNJJ-2014004)

摘　　要：针对传统蚁群算法及模糊C-均值聚类算法在合成孔径雷达遥感图像分割中精度低下和收敛速度较慢的问题,该文提出了一种改进的自适应阈值的蚁群及模糊C-均值聚类算法,实现对复杂合成孔径雷达图像进行分割。针对不同的合成孔径雷达图像,首先利用最大类间方差法获取最优阈值,通过最优阈值干预避免蚁群算法陷入局部最优解;再将自适应阈值蚁群算法得到的聚类中心和聚类类别数输入模糊C-均值聚类算法中,最终实现图像分割。实验结果证明,该算法在时间和误分率上较传统方法有显著的改进。An improved adaptive ant,combining with fuzzy C-means for complex synthetic aperture radar image segmentation was proposed for the disadvantages of traditional ant colony optimization and fuzzy C-means in image segmentation precision and convergence speed.According to different synthetic aperture radar images,the Otsu algorithm was used to obtain the optimal threshold,the optimal threshold intervention was used to avoid trapping in local optimal solution for ant algorithm;then the center and number of clustering which obtained by adaptive ant colony algorithm were input into fuzzy Cmeans.Experimental results showed that the proposed method was more efficient and precisely than traditional approaches in the segmentation of synthetic aperture radar images.

关 键 词：SAR图像分割 蚁群算法 模糊C-均值聚类 自适应 

分 类 号：TN957.52[电子电信—信号与信息处理]