Morphological segmentation of wear zones on cutting inserts for tool condition monitoring

This paper presents a low-cost, vision-based model for detecting and segmenting wear zones on cutting inserts. The system integrates a Vividia 2.0 MP handheld USB microscope with a MATLAB-based algorithm to enable automatic image segmentation. A total of 250 images were collected during machining on a CNC lathe (TCN 410-1250, SINUMERIK 840D), under controlled laboratory conditions, covering various wear states. The model includes pre-processing, morphological segmentation, visualization, and dimensional analysis. Results show high segmentation performance, with an accuracy of 0.99 ± 0.001 and precision of 0.92 ± 0.017. The proposed approach eliminates the need for expensive imaging setups by relying on a low-cost microscope and a simple computational pipeline. While the current setup was tested under controlled laboratory conditions, results indicate strong potential for reliable in situ monitoring of tool wear. The system is designed to be easy to implement and scalable, and this pilot study provides a solid foundation for future validation in real-world manufacturing environments. By supporting early detection of tool degradation, the approach could contribute to predictive maintenance strategies aimed at improving productivity, reducing downtime, and enhancing product quality.