Research On Gesture Recognition Based On Kinect And Its Application In Virtual Simulation System

Human-machine interaction technology is attracting more and more attention from researchers.The introduction of virtual reality technology has pushed the development of human-computer interaction technology to a new height.The combination of virtual reality and human-computer interaction technology is more and more applied in the fields of military simulation,medical rehabilitation,game creation and so on.Human action analysis is an important research topic in the field of human-computer interaction.The study of gesture recognition has the characteristics of convenience,intuition,strong interaction and rich expression of information.It has become the first choice for many researchers to analyze human behavior.However,traditional motion recognition is generally based on color images,but RGB images are easily interfered by background disturbance,light,environment and other factors.In recent years,more and more researchers begin to use the depth image to recognize human gesture.Based on this,we study the Microsoft Kinect device,which can get the depth image,RGB image and skeleton image of human body.We use the depth image and convert it into bone image and get spatial coordinate information of bone joint point.The main work of this article is as follows:1)This paper introduces the background and significance of the action recognition system,and then expounds the application of human behavior recognition in human-computer interaction and virtual reality technology.2)Secondly,in order to better understand the human action behavior,the hardware components of Kinect and the principle of Kinect for Windows SDK are introduced in detail.It also introduces the working principle of Kinect camera,the depth image and the acquisition principle of the skeleton image,and how to obtain the human skeleton information based on the depth information.3)Thirdly,we introduce the human action feature extraction method and propose different recognition methods for simple and complex actions.The spatial coordinate information of the joint points is used to carry on the simple action recognition,while the complex action is solved by the dynamic time warping.At the same time,it points out the shortcomings of the traditional dynamic time warping algorithm and optimizes it.The results show that the improved algorithm can improve the accuracy of gesture recognition.4)Finally,a virtual battlefield interactive environment is established through Unity 3D,which uses Kinect and mouse as dual channel.Besides,the recognition of human action gesture instruction and the control of the model of armored vehicle are realized.Finally,we show the achievement of synchronized movement between users and virtual armored vehicles based on Kinect.