Machine Learning Driven Image Analysis of Fine Grinded Knife Blade Surface Topographies

Abstract The optical perception of high precision, fine grinded surfaces plays a major role, especially in various consumer goods. The very complex manufacturing process of cutlery consists of a variety of parameters such as feed rate, cutting speed, grinding disc, cutting fluid, contact force or process time. The change of a parameter setting has a direct effect on the surface topography. Therefore, a standardized and optimized configuration of process parameters enables a desired quality of a product. In this study, the image analysis of the fine grinded surface is performed. For this reason, a customized test rig is designed in the first step. Here, constant boundary conditions such as defined light, specified camera settings, and a fixation ensuring the proper positioning of the samples are considered and determined for the further, computational analysis. The gathered image material serves as the training data for the machine learning analysis. The image of each grinded sample is analyzed concerning the measured roughness (Ra and Rz) and gloss value subjected to the corresponding production parameters based on Computer Vision (CV) techniques. Above all, Neural Networks as well as inductive methods are used for the image analysis. Additionally, the color data (L*a*b color space) is extracted using CV and serves as additional information for the further analysis. The application of the methodology is shown within the real case study.