HAND GESTURE RECOGNITION AND MOTION ESTIMATION USING THE KINECT SENSOR

摘要

This paper proposes and implements an efficient method to recognize hand gestures and estimate the motion of each fingertip using a combination of RGB image and depth image data acquired from Microsoft's Kinect sensor. The real-time performance of hand identification, tracking and motion estimation is achieved without any assistance of additional electronic devices. To guarantee the robust performance, differences between subsequent frames are analysed and specific algorithms are utilized in the proposed system. In addition, the calibration using both focal length and angle range of the camera is applied to coordinate with different devices to obtain accurate speed estimation. The experiment results show the robust and real-time performance. The proposed approach can be widely applied to robotic applications, such as visual servoing and human-robot interactions.