The influences of light source and roughness ranges on colour image-based visual roughness measurement performance

Abstract

This paper conducts a series of imaging experiments under different conditions to optimize the experimental parameters in the colour image-based roughness measurement method. The conditions consist of three colour blocks (a red-green (RG) colour block, red-blue (RB) colour block, and blue-green (BG) colour block), two roughness ranges and four light source brightness levels. The results show that the assessment effect of the HVS and the performance of the CDSM-based feature indices can be improved by using an RB colour block; the resolution of the HVS-based surface roughness evaluation method decreases with the increase in surface roughness; the energy index Ep can be used to measure a grinding surface with 0 < Ra ≤ 0.1 μm (100 nm) with a mean absolute error (MAE) of 10.7 nm; even under low energy field, the selected feature indices can be used to measure grinding surface roughness, which is significant for energy saving.