Research On The Application Of Hand Gesture Interaction Method Based On HTC Vive

With the development of virtual reality(VR),its interactive technology has become the focus of research and users.Today's virtual reality products are more focused on energy and investment,pay more attention to the display of visual effects,and more or less neglect of interactive means.Today's interactions are mainly through the use of some or traditional or specific electronic devices to interact,such as mouse,keyboard,handles,etc.,although the interaction problem is temporarily solved,but it also brings some problems.First,the user is extracted from the virtual world,and the immersion is greatly reduced.Second,the dependence on certain specific electronic devices is strong,resulting in high cost of money and learning.Under this condition,a little bit of gesture interaction appears.For humans,using gestures to interact is an instinct that can effectively reduce learning costs and increase immersion.Gesture interaction relies more on computer vision and computer graphics,and does not rely too much on specific electronic devices.This point of view uses gesture interaction to control the cost of money.Therefore,it has important research value for gesture interaction in virtual reality.The main research in this paper is to carry out the gesture interaction of immersion,accuracy and real-time in the VR head-up of HTC Vive.The main work includes the following points.A method of hand image fusion with virtual reality is proposed.In order to improve immersion and reduce the dependence on specific electronic devices such as handles and gloves,video information collected by monocular camera is used as input data to preprocess.In this step,several common skin color segmentation methods are compared and analyzed,and an improved adaptive threshold segmentation method is proposed to adapt to the application scenarios in this paper.The hand image is segmented from the complex background by this method and fused with the virtual environment.A method of gesture recognition in immersive VR is proposed.Because the data set of the first person gesture is very small,the data set for this experiment is established first,which is constructed by combining the existing open source data set with the data collected by the author.Because the direction of the first person gesture is not obvious,the histograms of oriented gradient(HOG)which has been widely used in pedestrian detection in recent years is selected to extract gesture features.Support Vector Machine(SVM)is used to classify and recognize gestures,and the size of data sets is analyzed and selected during training.Choose.Finally,the method is compared with Hu moment feature extraction method and Zernike moment method,which shows the advantages of this method in accuracy,and also can meet the real-time requirements.Set up a VR gesture interaction example platform to implement gesture interaction in HTC Vive.Experiments show that this method can meet the requirements of accuracy and real-time.