项目名称: 民航视景仿真的照片级逼真度渲染关键技术研究
项目编号: No.U1333110
项目类型: 联合基金项目
立项/批准年度: 2014
项目学科: 无线电电子学、电信技术
项目作者: 贺怀清
作者单位: 中国民航大学
项目金额: 36万元
中文摘要: 本课题研究民航训练仿真的关键问题:视景的照片级逼真度渲染。课题基于蒙特卡罗的光照传播计算,主要包括:(1)研究光照合成图象的滤波与重建,研究能够统一抑制像素间非一致性噪声和异常噪声的合成图象滤波器,利用相邻帧图象的相关性研究序列图象的滤波;(2)综合考虑像素抽样的随机性与相关性特点,研究评价像素噪声水平的量化指标,研究光照合成图象的自适应抽样;(3)研究Metropolis光照传播计算重要性抽样框架下的接收概率设计,研究合成图象平面上的光照变化与随机光径变化的函数关系、探讨Metropolis光照的本质和机理。研究成果不仅对民航训练仿真视景模拟具有重要的应用价值;在理论上,也促进了基于蒙特卡罗光照计算的照片级逼真度渲染技术的进步。
中文关键词: 视景仿真;照片级逼真度;滤波与重建;Metropolis重要性抽样;图像去噪
英文摘要: This research project is focused on rendering with high fidelity, used for visual simulation in the field of training for civil aviation. Our research is based on the simulation of light transportation by Monte Carlo. Mainly, the topics of this research project include: (1) filtering and reconstruction methods, considering the inter-pixel incoherence and spiky noise, for the synthetic images by Monte Carlo sampling; (2) quantitative evaluations on the noise level presented by pixel samples and how to perform adaptive sampling for the whole image synthesized; (3) the study of functional dependence between pixel value and sample path, and the acceptance probability design, in the framework of Metropolis light transport. The results achieved by this research project will largely benefit to the visual simulation for training in the field of civil aviation. Also importantly, the fruitful results obtained will improve significantly the Monte Carlo based light simulation.
英文关键词: visual simulation;high fidelity;filtering and reconstruction;Metropolis importance sampling;image denoising