Figures from this paperAbstractFigures from this paperAbstractFigures from this paperAbstractFigures from this paper

Abstract

： Received 3 January 2024 Accepted 11 March 2024 Available online 12 March 2024 With the rapid development of science, technology and electronic products, more and more intelligent products have complex structures and high integration levels. These products have extremely harsh electromagnetic environments due to factors such as mixed frequency bands, large power, and dense distribution. In order to solve the problem that millimeter-wave radar cannot effectively capture human body motion characteristics in an electromagnetic interference and noise environment, an anti-interference recognition algorithm based on two-stream feature fusion is proposed. The algorithm consists of three modules: distance feature extraction, Doppler feature extraction and feature fusion. The spatial pyramid pooling algorithm and channel attention mechanism are used to improve the VGG16 network and ResNet50 network respectively, improving the feature extraction capabilities of the network. Experimental results show that the recognition accuracy of our algorithm is 98.5%, and comparison with typical algorithms in the same field verifies the robustness of the algorithm in electromagnetic interference noise scenarios.