HRRP Open Set Target Recognition Algorithm Based on Decision Tree
Chen Maoyin, Chen Huaixin · 2024
High-Resolution Range Profile (HRRP) is a crucial radar signal processing technique widely employed in Radar Automatic Target Recognition (RATR). However, recognizing unknown targets under open-set conditions remains a significant challenge for HRRP. This study presents a decision tree-based open-set HRRP rapid target recognition algorithm, with two main contributions: first, four robust features are selected by comprehensively evaluating Fisher scores and the distribution characteristics of each feature; second, a reconstruction-based method is utilized to calculate the classification confidence of the decision tree's output, thereby effectively rejecting unknown targets. Experimental results demonstrate that the proposed method exhibits high recognition speed and robustness on measured datasets.