基于成像逼真度的视觉仿真方法

Visual Simulation and Evaluation Based on Imaging Fidelity

随着我国大型文艺表演活动次数的不断增加,在特定的仿真预演中,呈现人眼视觉效果的画面成为了迫切需求。文中根据当下图形仿真引擎的渲染管线处理画面的原理和流程,参考对于人眼视觉成像的相关研究,提出一种在基于物理渲染理论下,实时渲染中,人眼视觉画面的生成方法。在此方法上,对画面的组成部分(相机、光照和材质)进行了实验,通过和真实图像的对比,统计数据误差概率和权重,进行神经网络训练,从而评估最终的人眼视觉仿真结果。在国庆70周年群众游行仿真系统的设计和制作过程中,利用这一评估手段对摄像机画面进行视觉评估,为转播团队进行可信性验证,并对仿真效果进行评估,帮助决策者理解仿真结果,帮助制作团队快速找到仿真画面的不足之处,从而进行改正,取得了良好的实际效果。

With the continuous increase in the number of large-scale cultural performances in our country,in specific simulation rehearsals,images that present human visual effects have become an urgent need.Based on the principle and process of image processing by the rendering pipeline of the current graphics simulation engine,and referring to related research on human visual imaging,this paper proposes a method for generating human visual images in real-time rendering based on the theory of physical rendering.In this method,this paper conducts experiments on the components of the picture(camera,lighting and material),and performs neural network training through the comparison algorithm with the real image,statistical data error probability and weight,so as to evaluate the final human vision simulation result.During the design and production of the mass parade simulation system for the 70 th anniversary of the National Day,this evaluation method was used to visually evaluate the camera images,verify the credibility of the broadcast team,and evaluate the simulation results to help decision-makers understand the simulation As a result,it helped the production team to quickly find the shortcomings of the simulation images,and then corrected them,and achieved good actual results.