A System for the Quality Inspection of Surfaces of Watch Parts

Abstract

AbstractIn luxury industry and, in particular in watch making, the quality of a surface is fully associated with its visual appearance and represents a corollary to the technical mastery necessary to manufacture valuable products. Traditionally, the inspection of these surfaces is carried out by human experts. Their judgment is influenced by several factors, which are not easy to control and which introduce variability in quality inspection. Nevertheless, experts have the capacity to handle different situations and they only give access to the specific knowledge related to the inspection of aesthetic surfaces. For these reasons, in the development of systems for automated visual inspection, experts are considered as the reference. The main goal the work presented in this paper is to provide automated tools for the quantitative estimation of the quality of aesthetic surfaces, in order to reduce the variability of the inspection. The different parts of an artificial vision system for surface quality control are investigated and a parallel is drawn with the inspection carried out by human experts. The adopted inspection process is based on three steps: the identification of defects, their quantification and the judgment about their acceptability.KeywordsLinear Discriminant AnalysisHuman ExpertQuality InspectionInspection ProcessDirectional IlluminationThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.