Research On Human Action Recognition Algorithm Based On Ultra-wideband Radar

With the development of technology,human action recognition,as an important auxiliary technology in smart home,has significant research significance and broad application prospects in the fields of security monitoring,medical monitoring and human-computer interaction.Human action recognition technology based on ultra-wideband(UWB)radar has unmatched advantages over other recognition technologies,so this thesis studies the human action recognition based on UWB radar.In order to eliminate the interference of clutter and noise in radar signal on human action recognition and improve the accuracy of action recognition under small sample data,on the basis of removing the clutter and noise interference,this thesis proposes a human action recognition algorithm by UWB radar based on global and local features.Firstly,the original radar echo signal is preprocessed by moving target indication(MTI)combined with adaptive median filtering.Secondly,the principal component analysis(PCA)is used to extract the main components of the radar time-distance dimensional feature image of human action as the global feature representation,and the 2D discrete wavelet transform(2D-DWT)combined with singular value decomposition(SVD)is used to obtain the local feature representation of the feature image under different directions and scales,then fuses the global and local features.Finally,according to the fusion features,human action recognition is realized in the support vector machines(SVM)model.The research results show that the recognition accuracy of the algorithm in this thesis is94.17%,compared with the global and local features of actions,the accuracy is increased by2.09% and 3.34% respectively,and it has good recognition performance.In order to fully explore the action feature information in UWB radar signals,improve the parameter optimization speed of the SVM model and avoid the optimization falling into the local optimal solution,this thesis proposes a human action recognition algorithm by UWB radar based on time-frequency analysis.In the feature extraction stage,the preprocessed radar echo signal is subjected to a time-frequency transformation method based on signal energy proportional weighting to obtain the radar time-frequency feature image of human motion,then the kernel principal component analysis(KPCA)is used to extract the kernel principal components of the action time-frequency feature images as the nonlinear feature representation of human action.In the recognition and classification stage,human action recognition is realized in the support vector machines(SVM)model optimized by improved particle swarm algorithm(PSO)based on adaptive weights and mutation.The research results show that the recognition accuracy of the algorithm in this thesis is 96.25%,which is better than the compared algorithms,and the accuracy of human action recognition is effectively improved.There are 29 figures,11 tables and 69 references in this thesis.