Teaching Gesture Recognition Based On Depth Image Of Kinect

With the rapid development intelligent information technology,human-computer interaction(HCI)is extending toward more innovative and humanized.The traditional way of HCI depended on the hardware of equipment such as mouse,keyboard,handle for auxiliary interaction,which is not straightforward.Based on the advantage of flexibility,obvious and rich information expression,gesture interaction plays an important role in the field of HCI.Aiming at the problem of the traditional vision-based gesture recognition method can be fragile influenced by the light conditions and skin-like background environment.Based on the depth of information and skeletal joint data got form Kinect,the exploratory research of static gesture recognition and dynamic gesture recognition was set to work.The results of gesture recognition were applied to multimedia teaching system.In the static gesture recognition system,the depth information threshold value was set to select the area of interest from hand,the manpower was divided from the depth image by using the K-means.The Graham Scan algorithm was used to realize the gesture convex hull,and the defect points in the convex hull were detected.Combining convex hull with defect points can be more precise to found real fingertips.Identify digital static gestures based on the number of finger points detected,experiments show that the gesture recognition rate was higher,the validity of the algorithm was verified.Finally,finger fingertips on the basis of detection,achieving fingertip operation picture browsing.In the dynamic gesture recognition system,based on the human skeletal joint data obtained by Kinect sensor,the four key position information of hand useful for the input of hand eigenvector was centralized and normalized.In this article,Dynamic Time Warping(DTW)algorithm was applied to identify dynamic gestures.The DTW was optimized through global path limit and distortion threshold,as a result,the real-time performance of DTW was improved.At last,experiments verify that the DTW effectively identify the dynamic gesture.Meanwhile,DTW algorithm was verified in the strong and weak light conditions.The results of dynamic gesture recognition was applied the PPT operating system which can control PPT play without touch.