Research And Implementation Of The Technology Of Large Scale Scene Management And Wondering

Large scale scene management and wondering technologies have been playing the important part among various researches in the field of virtual reality. Although the graphics hardware's processing ability has improved a lot, it can not afford thousands of objects in the scene efficiently. The thesis' objective is to design and implement a real-time large scale scene wondering prototype system, concentrating on the subjects of scene management and terrain in the process, so as to enhance its Real Time Capabilities.The thesis first introduces the architecture of real-time large scale scene wondering system, and then analyzes the main function and the relationship with others of the scene management module. Based on these researches, we design and implement three main system modules: rendering module, resource management module and scene management module. Aiming at the large scale terrain data, we introduce the resource pool mechanism into the scheduler module, enhance the real time capabilities of the system. Moreover, based on the idea of transferring the load from CPU to GPU, we implement a quad-tree-based terrain patch management structure, and the LOD algorithms.