Investigations on electromagnetic wave scattering simulation from rough surface: Some instructions for surface roughness measurement based on machine vison

Abstract

There are many experimental variables in the research of surface roughness measurement based on machine vision technology, such as light incident angle, light wavelength, workpiece material, surface features. It is, however, time-consuming and also labour-intensive to conduct experiments considering all these variables. In this work, we proposed to conduct Electromagnetic Wave Scattering Simulation considering these variables using finite element method (FEM) in order to give instructions for actual imaging design. The simulation process is more theoretically robust, more efficient and cost-effective rather than actual experiments with empirical models. In the simulation process, a simulation model is designed first. In this model, the light incident angles θ include 15°, 30°, 45°, 60° and 75°; the light wavelengths λ include 400 nm, 550 nm and 700 nm corresponding to blue, green and red; the workpiece materials include stainless steel (SUS), copper (Cu), aluminium (Al), and silicon carbide (SiC). Then, a series of rough surface with different surface roughness values are obtained by defining cosine functions and non-Gaussian surfaces. The simulation results show that given material, when 0≤Ra ≤ 0.1 μm, the performance of small θ is obviously better than that of large θ. When 0.1≤Ra ≤ 1 μm, large θ is better (θ ≥ 60°). This derives to obtain better measurement accuracy.