Research Of 3D Gesture Data Processing Technology For Human-machine Interaction

In recent years, Thanks to the development of visual sensor and computer vision technology, vision-based gesture interaction technology has achieved a great progress.This paper carries out research of 3D gesture data processing technology on the basis of the human-machine interaction system developed by Intelligent Robot Lab. The human-machine interaction system adopts 5 Leap Motion sensors as the input devices of the system, and captures position data and posture data of user’s hand without auxiliary contact-type sensors or color marks. In this paper, the main work is as follows:1. Put forward a precise registration technology which combines the algorithm based onunit quaternion and UKF algorithm to achieve coordinate conversion between thelocal coordinate system of each sensor and the global coordinate system of thehuman-machine interaction system;2. Considering the random noise in position data of hand gesture captured by sensors,this paper adopts a kind of improved adaptive Kalman filter based on the ’current’statistical model to filter the random noise and fuse position data from differentsensors, and improve the accuracy of filter through the algorithm’s improvement;3. For the random noise in posture data of hand gesture, considering the non-linearrelationship between the sensor’s measured data and posture variable in system, theUKF algorithm is used to filter the random noise and fuse posture data fromdifferent sensors, and improve the stability and accuracy of posture data.In the last chapter of this paper, the algorithm and technology used in this system is verified by the experiment of Manipulator control.