Gesture recognition based on Kinect depth information
Zhang Qi-zhi · Journal of Beijing Information Science & Technology University · 2013
An algorithm of moving gesture recognition based on the depth information of Kinect and dual threshold segmentation is designed.The algorithm is realized under VS2010 by integrating OpenCV and OpenNI.Firstly,the depth camera of Kinect is used to get the depth image.Secondly,the dual thresholds are used to segment the image to get the image of just hand.Then morphology process is performed for the image to achieve more integrity hand shape.Finally,the GestureGenerator of OpenNI is utilized to track and recognize the gesture.In this method,the obstructions of background and part of foreground are removed by dual threshold segmentation.Points,circles and lines in different colors are used to describe the gestures,so the results of recognition can be clearly shown in real time.