局部线性嵌入(LLE)是一种有代表性的流形学习算法,利用核技术将LLE进行推广,得到核局部线性嵌入算法,并将其应用于雷达目标一维距离像的特征提取。然后采用一种基于核的非线性分类器,对所提取的特征进行分类。对3种飞机的实测数据进行识别实验,结果表明,该方法具有较优的识别性能。

Locally linear embedding(LLE)is one of the representative manifold learning algorithms.In this paper,LLE is extended using kernel technique,which leads to kernel locally linear embedding(KLLE)algorithm.KLLE is first utilized to ex- tract nonlinear features from a range profile.Then,a kernel-based nonlinear classifier,called KNR(Kernel-based Nonlinear Repr- esentor),is introduced and employed for classification.Experimental results on measured profiles from three aircrafts show rela- tively good recognition performance of this method.

TN957.5

教育部科学技术研究项目，ATR重点实验室基金