The Design And Implementation Of A Novel NET Confrontational Virtual Simulation System Based On Multi-view Stereo Display

The popular somatic game systems such as tennis game based on Kinect only provide a single view image to all players,which means they all see a same image when different players view a same screen display.However,compared with the real world,they should see different images due to different viewpoints.Sometimes,split-screen display technology is adopted for providing two images based on the perspectives of two players on the right part and the left part of a screen.However,the two images will interfere with each other,and stereoscopic display can't be supported in this case,which can't help players to get a better immersion and game experience.In this paper,we design and implement a novel net confrontational virtual simulation system based on multi-view stereo display and give an application in tennis game system.This system can break through the limits of above two display modes(single-screen display and split-screen display),and can also provide co-location users with individual views according to their position on a shared display.Besides,our system adopts stereo display technology to make the game more realistic.Players can control the corresponding virtual roles to act such as running or hitting the ball by real and natural motions,just like they do in real world,which can make the game more interactive and immersive.Our system mainly consists of four modules:game logic module,interaction control module;multi-view stereoscopic rendering module in real time(stereoscopic rendering module)and multi-view stereoscopic projection display module(multi-view projection display module).Game logic module is responsible for execution of game rules of the tennis match and input data processing,and also realize the functions of animation control,position mapping,physics engine and so on.Interaction control module use Kinect to achieve natural interaction.As the major equipment of interaction control module,Kinect is used to capture the positions of different players,identify their actions and send the input information to game logic module.The core modules are stereoscopic rendering module and multi-view projection display module.The position data of players captured by interaction control module was reasonably mapped to the corresponding position of virtual game roles in the game scene in real time by game logic module.Then,stereoscopic rendering module employs binocular camera equipped on virtual avatars to shoot the left and right frame images of virtual avatars' views.With the support of the NVIDIA 3D Vision driver,the left and right frames are rendered in GeForce graphic cards and get the stereo images of viewpoints of virtual roles.Meanwhile,stereo videos are hybrid coded and corrected by GPU in real time.Multi-view projection display module is composed of a projection metal screen,three DLP projectors and three kinds of 3D shutter glasses.These projectors and glasses needed to be modified to achieve the three stereo videos projected and viewed correctly and individually.The three projectors project the rendered and processed three stereo videos to a same display,so the three players wearing modified shutter glasses can view different stereoscopic images on a shared display.