基于直觉模糊c均值聚类核匹配追踪的弹道中段目标识别方法

针对核匹配追踪算法(KMP, kernel matching pursuit)进行全局最优搜索导致学习时间过长这一缺陷,汲取直觉模糊c均值聚类(IFCM, intuitionistic fuzzy c-means)算法的动态聚类特性优势,提出一种基于直觉模糊c均值聚类的核匹配追踪(IFCM-KMP, intuitionistic fuzzy c-means kernel matching pursuit)算法,且对UCI库中4组实际样本数据进行了分类实验及有效性测试.最后,选取高分辨距离像(HRRP)这一弹道中段目标识别常用的特征属性,对其进行特征提取获得子像,并分别采用FCM,KMP,IFCM-KMP 3种算法对真弹头进行目标识别仿真实验及结果对比分析,充分表明了IFCM-KMP算法用于弹道中段目标识别较之FCM、KMP的优越性及有效性.

Technique for target recognition based on intuitionistic fuzzy c-means clustering and kernel matching pursuit

Kernel matching pursuit requires every step of searching process be global optimal searching in the redundant dictionary of function.Namely,the dictionary learning time of KMP was too long.To the above drawbacks,a novel technique for KMP based on IFCM was proposed to substitute local searching for global searching by the property superiority of dynamic clustering performance,which was also the superiority in Intuitionistic fuzzy c-means algorithm.Then two testing including classification and effectiveness were carried out towards four real sample data.Subsequently,high resolution range profile (HRRP)was selected from the classical properties of target recognition in e middle ballistic trajectory,which were extracted for getting sub-range profile.Finally,three algorithms including FCM,KMP,IFCM-KMP were carried out respectively towards different kinds of sub-range profile samples in emulation platform,the conclusion of which fully demonstrates that the IFCM-KMP algorithm is superior over FCM and KMP when it comes to target recognition in the middle ballistic trajectory.