HRRP Clutter Rejection Via One-Class Classifier With Hausdorff Distance

Abstract

Lots of detectors for high resolution range profile (HRRP) data are mainly based on the intensities of target echoes. When the intensities of outliers are as large as those of targets of interest, some false alarms will emerge during the detection stage. In this article, we propose a rejection algorithm for HRRP data between the detection stage and recognition stage to eliminate the false alarms occurring during the detection stage. In this method, the intensities and positions of the dominant scatterers are extracted as a joint feature. Then, the feature is used to develop the K-center one-class classifier based on Hausdorff distance. Experimental results on the measured HRRP data indicate that the proposed rejection algorithm can eliminate the false alarms remarkably well, and keep most of the target data as well. Meanwhile, the proposed method has better performance under different values of signal-to-noise ratio and different values of parameters than some other rejection methods.