Research On The Key Technologies Of Real Time Virtual Display For Complex Object

With the rapid development of computer technology, virtual reality technology is used in more and more areas. In the meantime, it needs to generate realistic images in real-time. Especially for application of complex objects, which have enormous data, generating realistic images in real-time has been the bottleneck and key point in virtual reality application areas. In virtual reality system with complex objects, the realistic degree of complex objects can affect users' immersion directly.In this dissertation, we mainly focus on real-time realistic rendering of the virtual complex objects. Our major research work includes:1) The current problems and future direction of real-time virtual display are given after analyzing the state-of-art of real-time virtual display and related technologies.2) The method on real-time processing of point cloud data is presented, which is based on CUDA programming model of NVIDIA. It can efficiently utilize the parallel power of GPU and improve the efficiencies of real-time processing of huge point cloud data. The data structure generated by this method can be used to speed up real-time realistic rendering.3) The real-time rendering technology is discussed ,which is based on multi-core and screen space and uses hierarchical space data structure to dynamically compute the set of sample points, instead of randomly generate the set of sample points. The hierarchical data space structure can simplify the choose of sample point, reduce the time of computation, save the memory. In the meantime, this technology can implement multi-resolution rendering with view-dependent LOD control. Since the hierarchical space data structure based on Octree is deployed to represent complex object, which can be generated by CUDA on the fly, the algorithm adopted by the technology not only speeds up the real-time rendering but supports the dynamic complex objects.4) The hybrid rendering technology based on PRT and BRDFLW(Bi-Directional Reflection Distributed Function with Light Weight) is investigated. It combines the traditional BRDF and the set of point light from HDR irradiance map, and calculates the weight coefficient of set of point light. Then it integrates weight coefficient into light model based on BRDF. Since this technology computes the effect of set of point instead of point light itself, the speed of rendering is improved. It adopts rendering method based on PRT in order to generate shadow, so the excellent image can be acquired.5) The research on how to render transparent object with dynamic HDR background light is made to solve the effect of light such as scattering, reflection, refraction of transparent material.6) We implement the tone mapping system based on weight coefficient of luminance. The weight coefficient of every pixel is accounted by light distribution of this pixel in HDR irradiance map. This system not only reserves local detail but reduce the loss of image information.Based on the aforementioned core technologies, we also develop and implement an interactive virtual display system, which mainly include car driving subsystem, editor subsystem of material based on GPU and display subsystem of complex product (car).The display system of complex products can real-time reflects the effect of light and shadow of inner and outer part of car when the light of HDR background map is changed dynamically.