Online detection technology of triangular-blade tool grinding precision based on machine vision

Abstract

To address the problems of repeat positioning accuracy and secondary clamping caused by a triangular-blade tool during grinding, this paper proposes an online detection method for the machining accuracy of the triangular-blade tool based on machine vision. The method utilizes a weighted average approach for grayscale processing of the original image and an adaptive median filtering algorithm for filtering and noise reduction. The processed image is then binarized. The Canny and Zernike moment edge detection algorithms are utilized for pixel-level and sub-pixel-level edge positioning. A curvature-based feature extraction method is proposed to complete image stitching. Measurement software is designed and developed on the MATLAB app designer platform. Experimental results show that the relative error in tool grinding length is within 0.094%, the average error in cutting edge width is 1.105%, the relative error in thickness is 5.065%, and the relative error in symmetry is 6.044%. The accuracy of the proposed method is confirmed through a comparison between image and microscope measurement.