False‐alarm suppression with random forest by exploiting ambiguity features of targets. Issue 24 (14th October 2022)

• Record Type: Journal Article
• Authors: Wang, Zhifei; Yu, Junpeng; Yang, Yuhao
• Is Part Of: Electronics letters. Volume 58:Issue 24(2022)
• Page Start: 917
• Page End: 919
• Publication Date: 2022-10-14
• Subjects: Electronics -- Periodicals; 621.381
• DOI: 10.1049/ell2.12642
• Languages: English
• ISSNs: 0013-5194

Abstract

Abstract: Suppressing false alarms without impacting detection rate is an essential issue for search radars. Traditional threshold detection methods identify true targets by simply comparing the energy of target candidates with a threshold. In this work, a new data‐driven false‐alarm suppression approach is proposed based on a random forest model. It discriminates true targets and false alarms in the developed multi‐dimensional ambiguity feature spaces, instead of the time‐frequency features used in previous works, to fulfill short dwell‐time requirement of search radars. Evaluations through field experiments demonstrate that the proposed method can achieve as high as 98% validation accuracy and a significant improvement of detection performance compared to the conventional threshold detection.