Abstract
AbstractSurface roughness is an essential property in the manufacturing industry to assess the quality of its products after finishing operations. However, the evaluation of surface quality in wood products usually depends on the subjective inspection of the operators, which implies a high variability in the final quality of the pieces. This study proposes a new method to estimate roughness parameters by applying algorithms on images of wood parts processed by robotic sanding. For this purpose, this article presents a hybrid approach based on features using the co-occurrence matrix applied to greyscale images processed with five edge detection algorithms. For the evaluation of the performance of this method, the researchers correlated five features for each edge detection algorithm with standard surface roughness parameters, obtaining high correlations. The results of this study constitute a first step in implementing the proposed method in inspection systems for optical roughness measurement of wood products in automated industrial environments.
Publisher
Springer International Publishing