Research And Implementation Of Gesture Recognition And Collision Detection In Virtual Reality Interaction

Augmented Reality(AR)technology is mainly used to superimpose virtual objects generated by computer technology onto the real environment.In addition,the integrated scene is presented to the user,which can superimpose richer information for the real scene in the real world,so as to improve the sense of immersion of the system and bring more real interactive experience to the user.Augmented reality interaction makes the use of external equipment to obtain the corresponding image,audio,sensor data and other corresponding information as data,through the data information to make a series of responses to realize the control of virtual objects,so as to achieve the purpose of human interaction with virtual objects.The traditional enhanced interaction mainly adopts data gloves,Holo Lens,Kinect,Leap Motion and other hardware devices.Although it is found in the interaction that the user has a high sense of experience in the process of use,the promotion is greatly limited due to the high cost of motion sensing equipment.Therefore,more and more scholars have attracted attention.Through the study of the existing interaction theories and methods of augmented reality,it is found that in the traditional interaction scenes of augmented reality based on monocular cameras,interaction immersion is poor,the sense of substitution is not strong,and the interaction mode is not natural and smooth.Therefore,this thesis proposes a method of gesture recognition and collision detection in virtual-real interaction based on monocular camera to solve the above problems.Not only improves the sense of immersion and reality of augmented reality interaction,but also improves the fluency and portability of the augmented reality interaction system,so that the virtual and real interaction system can be popularized at home and abroad.The main work of this thesis is as follows:(1)Gesture recognition based on monocular camera is studied,and gesture recognition is completed by deep learning method to ensure the accuracy and real-time of gesture recognition.(2)Six kinds of gesture instruction information are designed.Gesture interaction technology achieves the control between human and virtual objects.And this provides more interactive possibilities.(3)Due to the diverse and complex environments in the real world,problems such as poor sense of reality and weak sense of substitution are found in complex scenes.In this thesis,point cloud reconstruction and octree collision detection methods are adopted to achieve the effect of virtual-real collision and improve the sense of reality and immersion of the augmented reality system.(4)Combining the methods of gesture recognition and collision detection,this thesis designs a set of virtual-real interaction system.A complex virtual object is selected,and users can control the virtual object and realize the virtual-real collision effect through gestures,which solves the problems of poor immersion and non-smooth interaction in augmented reality interaction.