Design and Implementation of an Online Machine Vision Dimensional Inspection System for Precision Cutting Tools

Abstract

Traditional offline measurement methods in precision tool manufacturing suffer from low efficiency and lack of real-time feedback. This study developed a machine vision-based online dimensional inspection system integrated into a five-axis grinding machine. The system employs a CCD industrial camera with a telecentric lens. A dual-mode illumination scheme combining ring lighting and coaxial telecentric backlighting was adopted to capture images of the tool end face and peripheral cutting edges. To address the challenge of multiple imaging planes during tool rotation, a multi-frame image fusion algorithm was proposed. Tool positioning and measurement alignment were realized using contour-based template matching and rigid transformation, supplemented by sub-pixel edge fitting for improved accuracy. Closed-loop feedback was achieved through an MES-fog computing architecture, enabling automatic adjustment of the machining program. Experimental results demonstrate that the system attains a pixel resolution of 0.0031 mm, with standard deviations of the six key dimensional measurements for four-flute end mills all below 0.006 mm. The system has been preliminarily deployed on an actual production line.