Description |
Three-dimensional (3D) models of industrial plant primitives are used extensively in modern asset design, management, and visualization systems. Such systems allow users to efficiently perform tasks in Computer Aided Design (CAD), life-cycle management, construction progress monitoring, virtual reality training, marketing walk-throughs, or other visualization. Thus, capturing industrial plant models has correspondingly become a rapidly growing industry. The purpose of this research was to demonstrate an efficient way to ascertain physical model parameters of reflectance properties of industrial plant primitives for use in CAD and 3D modeling visualization systems. The first part of this research outlines the sources of error corresponding to 3D models created from Light Detection and Ranging (LiDAR) point clouds. Fourier analysis exposes the error due to a LiDAR system's finite sampling rate. Taylor expansion illustrates the errors associated with linearization due to flat polygonal surfaces. Finally, a statistical analysis of the error associated with LiDar scanner hardware is presented. The second part of this research demonstrates a method for determining Phong specular and Oren-Nayar diffuse reflectance parameters for modeling and rendering pipes, the most ubiquitous form of industrial plant primitives. For specular reflectance, the Phong model is used. Estimates of specular and diffuse parameters of two ideal cylinders and one measured cylinder using brightness data acquired from a LiDAR scanner are presented. The estimated reflectance model of the measured cylinder has a mean relative error of 2.88% and a standard deviation of relative error of 4.0%. The final part of this research describes a method for determining specular, diffuse and color material properties and applies the method to seven pipes from an industrial plant. The colorless specular and diffuse properties were estimated by numerically inverting LiDAR brightness data. The color ambient and diffuse properties are estimated using k-means clustering. The colorless properties yielded estimated brightness values that are within an RMS of 3.4% with a maximum of 7.0% and a minimum of 1.6%. The estimated color properties effected an RMS residual of 13.2% with a maximum of 20.3% and a minimum of 9.1%. |