Structural Feature Representation and Fusion of Human Spatial Cooperative Motion for Action Recognition

Aiming at the cooperative relationship of human body parts in the process of action execution, we propose an action recognition method based on the structural feature model of human spatial cooperative motion. First, ten wearable sensors and Kinect v2 are used to collect human motion data. Second, we analyze the relationship between the three-axis acceleration data of multiple sensors. Third, we measure the contribution of different parts of the human body to the completion of movement. And the contribution of different parts is transformed into the structural feature model of cooperative motion. Finally, we apply unsupervised and adaptive constraints to the motion features of different parts of the human body. On this basis, the features of different modals are fused. The experimental results show that our method can significantly improve the recognition rate of the open test. At the same time, the calculation process of our method is simple and easy to implement.