Kinect-based Hand Gestrue And Face Recognition And Its Application On Education Area

With the development of computer and related fields,people's requirements on interactive experience are getting higher and higher.People's hands and faces contains much information,so hand gesture recognition and facial recognition has become a hot topic this years.Kinect is a high-tech and inexpensive device which could provide both depth information and skeleton flow information.It's a significant advantage for hand gesture recognition,3D facial recognition and applications on these area.There are three parts in this paper.On hand gesture,we studied the current main method of hand segmentation and gesture recognition algorithm,especially focus on recognition algorithm based-on template matching.We studied dynamic time warping(DTW)algorithm and proposed an algorithm to improve the accuracy and efficiency on hand gesture recognition.On facial recognition,we focus on algorithms which relate facial feature extraction.Then,we proposed a face localization algorithm called Skeleton-Nose Localization.Meanwhile,we compared two feature extraction method,Local Binary Pattern(LBP)and Active Appearance Model(AAM).On application,we developed two Kinectbased application,the one is a gesture-control 3D globe scene,the other is a facial-recognition-based learning condition monitoring system.The algorithms we proposed in this paper on hand gesture recognition and facial localization is a certain innovation.Two application we developed based on the research in this paper is both innovative and useful.Although we choose Kinect as the main tool for research in this article,the results of this paper and its prospects are not confined to Kinect.