Algorithm Research For Global Occlusion Map And Exach PVS Based On View Cells

With continual development of technology such as virtual reality and 3D interactive application, large model's real-time rendering has become researching hotspot in computer graphics. As main accelerating method for interactive walkthrough in virtual scene, occlusion culling technology has been increasingly concerned. However, because of the scale and complex's increase of virtual scene, a great time has been used in preprocess phase. How to improve the preprocess speed in from-region occlusion culling technology has been a problem which should been solved urgently.Firstly, the algorithm of Global Occlusion Map was improved. For accelerating the GOM constructed process, through researching visible sets'relativity between adjacent view cells', the view cell's dependent GOM theorem was deduced. And, the GOM constructing method was designed based on using directed joint graph to express the dependence between view cells.Secondly, the exact PVS algorithm based on conservative occlusion culling method was designed to aim at traditional exact PVS algorithm. There is three aspects for accelerating the PVS computing process: above all, by use of the feature of space consistency, conservative occlusion culling method based on virtual occluders was used to quickly estimate mass invisible objects in scene; next, ray casting technique was utilized to confirm visible polygons; at last, occluders was optimized to improve polygon-polygon visibility method for distinguishing the uncertain polygons.Finally, the improved GOM algorithm and exact PVS algorithm based on virtual occluders were validated through tests. Opengl andVC++ was used to implement the algorithms, but also, performances of the traditional and improved algorithm was compared and analyzed.The preprocessing speed was efficiently increased through improved traditional GOM algorithm and exact PVS algorithm, consequently its'practical value was enhanced in scene's real-time rendering.