Radar Target Recognition Method Using Improved Support Vector Machines Based on Polarized HRRPs

Abstract

Target recognition based on high range resolution (HRR) polarized radar using support vector machines (SVMs) was studied in this paper. A fuzzy membership function was constructed based on SVM decision-making function in order to improve the performance of OAA and OAO classifiers for multi-class target, and HRR radar target recognition method using improved SVM was proposed: First, the polarized radar backscatter echoes were processed by incoherent integration and power-normalized, the location and length of target in echoes were estimated and range profiles of target were interpolated to certain radial length, then polarized profiles were integrated considering the relevancy of range profiles of same target in different polarization state, at last, the improved OAA and OAO classifiers were used for target classification. Simulation experiment results show that the proposed method has the advantage of little capacity of computation and can improve the performance of classifiers effectively.