Posture Recognition with Background Noise Elimination Using FMCW Radar

With the development of social intelligence and information technology, the research on human posture recognition has received extensive attention. In this paper, Posture recognition with background noise elimination using frequency modulated continuous wave (FMCW) radar. Firstly, the radar signal processing is processed by using the method of limiting the range of motion, which effectively eliminates the influence of some static objects. Secondly, Density-Based spatial clustering of applications with noise (DBSCAN) method is adopted to cluster the processed coordinate data onto different clustering groups, so as to eliminate the influence of dynamic and static objects. Finally, the multi-branch deep learning network structure is adopted to effectively eliminate the background noise in different environments. The experimental results based on the collected actual data sets show that the accuracy of human posture recognition is significantly improved.