基于改进ISODATA的无先验扩展目标聚类分析算法

Clustering Analysis Algorithm of Extended Targets without Prior Information Based on Improved ISODATA

基于弹载雷达多扩展目标检测的应用需求,在CFAR检测输出的基础上,对检测结果的聚类分析方法进行了论证分析,提出了改进的ISODATA算法。该算法摆脱了常用聚类分析算法对目标个数、聚类初值等先验信息的要求以及孤立点的影响。数字仿真实验验证了新算法的有效性。

Based on CFAR detection output,an improved iterative self-organizing data analysis（ ISODATA） algorithm is presented for multi-extended targets detection requirements of missile-borne radar. This algorithm can get rid of the influence of common clustering analysis algorithm on the prior information requirements,such as target number,clustering initial value,which can eliminate the effect on isolated points. Simulation results verify the algorithm effectiveness.