Research On3D Modeling And Transformation Of The Personalized Hand

With the rapid development of the computer age, more and more people pursuit theintelligent industry, so the man-machine combination and interaction are more reused. As acommunication method rank second only to the language, people are able to complete themachine instruction guide work through gestures,body gesture based on human-computerinteraction applications are also emerged in an endless stream, for example it has theextremely vital significance in sign language comprehension, virtual reality, robot control,entertainment and other fields. This kind of interaction focus on the artificial center changeswith harmonious and natural,there is a significant increase on the human-computerinteraction in natural and efficiency. The main research contents of this paper focus on themodeling of3D hand and its diversity, so we can better understand the diversitymechanism of gesture interaction.In the hands of3D modeling: firstly according to the complexity of the human hand,using Kinect to collect the3D point cloud from different angles, all point cloud under thedifferent perspective need to registration, streamlined, repair to obtain complete hand pointcloud. In the reconstruction process of point cloud surface reconstruction method isproposed based on the differential manifold, similar to the way of radial basis functions areused in the basic function of differential manifold construction, with the geodesic distanceinstead of the shortest distance between two points, and the adjacent patches of smoothstitching up through the neighborhood transition mapping. The experimental results showthat smoothness and continuity of the geodesic based on differential Manifold surface areexcellent. In the aspect of hand diversity: according to the physiological structure of humanhand,the research based on joint transform of gestures is put forward. Cross dimensionalmatching method for3D point cloud and2D image combination are used in the extractio nprocess of joints, and obtained new gestures through the three-dimensional coordinatetransformation according to law by the hand motion, experimental results show that thetransform is ideal. According to the characteristics of joint points of contour transform andacquisition gesture, gesture recognition is proposed on closure area based on image. Thematching results show, in the range of allowable error, gesture transformation is successful,consistent with the gesture shapes.