A New Radar High-Resolution Range Profile Recognition Method of the Maximal Margin Kernel Optimization

The radar automatic target recognition (RATR) base on the radar target High-Resolution Range Profile (HRRP) plays an important role in military and civilian field. But the target aspect sensitivity and high feature dimensionality of the target HRRP cause the nonlinear separability of the HRRP. Aiming at this problem, this paper proposes a new radar target HRRP recognition method based on the maximal margin kernel optimization. At first, the maximal margin principle is used for the data-dependent kernel optimization, and then the Support Vector Machine (SVM) is applied for the radar target HRRP recognition as a classifier. At last, the recognition simulation experiment based on 5 kinds of aircraft targets HRRP has been done. And the experimental results show that the proposed method based on the maximal margin kernel optimization can effectively optimize the kernel parameters considering SVM classifier, and thus improve the recognition performance.