Abstract:

In the development of interactive application systems based on the Microsoft Kinect somatosensory equipment,it is difficult to achieve desired results using traditional human computer interaction devices like mouse and keyboard.A gesture identification method based on fingertip detection is presented to address this problem.The depth information of the image is obtained using the Kinect sensor,and the position information of the palm center using the built-in module of OpenNI SDK.Then the nearest neighbor method is used to achieve hand segmentation and extract the hand shape.After that,the convex hull is obtained using the Graham Scan algorithm optimized with a ray intersection method.Finally,the fingertips are detected using the profile analysis method from the convex hull.On this basis,the recognition of the "grab" gestures is realized using the number of fingertips and the area ratio.Experiments results show that this method can effectively identify the grasp gesture and has good robustness.

Key words: fingertip detection;Kinect;OpenNI;gesture identification;human-computer interaction