Machine vision-based dimension measurement system for the end cap parts

Abstract

In the traditional machinery industry, dimensional inspection of parts mainly relies on manual operation, which suffers from low accuracy and low efficiency. This mode of inspection is unable to satisfy the demands for efficient and precise dimensional measurement.The application of machine vision has opened up a new approach to dimensional inspection of parts.Based on the MATLAB software development platform and using 2D images of parts captured by the industrial camera, this paper proposes an efficient measurement algorithm that determines the hole diameter dimensions of end cap parts through a series of image processing steps. Zhang's camera calibration method is adopted to establish the imaging model, and lens distortion is corrected via distortion parameter algorithms to enhance measurement accuracy. Preprocessing technologies such as denoising and filtering are adopted to enhance and optimize image quality. The outer contour of the hole is extracted using edge detection operators, and the Hough circle transform is subsequently applied for hole localization, yielding the center coordinates and radius. Finally, the actual diameter is obtained by combining the camera calibration parameters, thereby achieving precise dimensional inspection of the end cap hole diameter. Through comparative analysis of multiple detection algorithms, the measured absolute error of hole diameter is less than 0.04 mm, which meets the tolerance requirements for mechanical parts. The proposed method is highly applicable to the automated dimensional measurement of end cap hole diameters in industrial machining scenarios.