A HRRP Recognition Method Based on KFD

High resolution range profiles (HRRP) could accurately reflect the structure of target, so it is an important method for radar target recognition. Kernel Fisher discriminant (KFD), which is a machine learning method based on kernel function, is suitable for classification of high dimensional samples which couldn´t be separated by linear classifier. In this paper, KFD were used for HRRP classification with KMOD kernel function. A multiple classifier was proposed, and better anti-noise performance was achieved with phase-subtraction alignment and a special rejecting method. The experimental results by three classes of measured HRRP data proved out the effectiveness of KFD