Technique for Target Recognition Based on Intuitionistic Fuzzy C-Means Clustering and Kernel Matching Pursuit

Yang LEI,Wei-wei KONG,Ying-jie LEI

2012-01-01

Abstract: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.