Monte Carlo simulation of three-dimensional measurements of translucent objects

Abstract

In a previous article, we described the deviation between the real and the measured object surface that occurs when a translucent object is scanned by an active triangulation system. This error depends on the angle between the measurement direction and the object’s surface normal, the surface reflection behavior, which can be described by a bidirectional reflectance distribution function, and the light penetration behavior. In general, the error is small if the surface is perpendicular to the measurement direction; it increases if the surface is tilted and decreases again for flat angles. This error curve is additionally affected by the surface roughness. The angle dependence is more distinct for smooth surfaces. In order to predict and compensate for the error, it is necessary to understand the error-forming process. Therefore, Monte Carlo simulations of several measurements were performed. As the computational cost is very high for three-dimensional simulations, most of the simulations were performed in two-dimensional space. We present the results of these simulations and discuss how the measurement error depends on the surface roughness, the measurement direction, and the scattering behavior of the material.