Open AccessResearch on radar clutter recognition method based on LSTM
Author(s) -
Song Li,
Shengli Wang,
Dingbao Xie
Publication year - 2019
Publication title -
the journal of engineering
Language(s) - English
Resource type - Journals
ISSN - 2051-3305
DOI - 10.1049/joe.2019.0242
Subject(s) - clutter , computer science , artificial intelligence , radar , artificial neural network , constant false alarm rate , tracking (education) , pattern recognition (psychology) , computer vision , telecommunications , psychology , pedagogy
Considering the increasingly complex electromagnetic environment of radar detection, it still contains a lot of clutter after target detection, which is a challenging problem for subsequent target tracking. A radar clutter recognition method based on long short‐term memory (LSTM) neural network is proposed here. This method can further identify clutter points after target detection and improve the quality of target tracking. LSTM neural network is trained and compared with K nearest neighbour method and support vector machine. It is found that the clutter recognition accuracy of this method is up to 86.3%, which is 12.2 and 4.1% higher than the latter two methods. At the same time, the target loss rate of this method is only 16.6%, which is 18.3 and 5.7% lower than the latter two methods respectively. The experimental results show that the radar clutter recognition method based on LSTM neural network is effective.