Abstract: Aiming at the problems of insufficient precision, low efficiency and strong manual dependence in the measurement of key dimensions of stud bolts for high-speed railways, a method for measuring key dimensions of stud bolts based on machine vision is proposed. Image pyramid technology and improved fast non-local means filtering technology are combined for image preprocessing, which effectively suppresses burr interference and improves the extraction accuracy of fillet features. A four-directional Sobel operator was introduced to achieve accurate extraction of thread contour edges, and the least square method for linear fitting and circle fitting algorithms were used to measure the geometric parameters and chamfer features of stud bolts, respectively. Experimental results show that the method has significant measurement accuracy and stability. Compared with the comparison algorithm, the relative error of external thread major diameter measurement is reduced to 0.17%, and the standard deviation of bolt total length measurement is reduced to 0.006 8 mm, realizing high-precision and high-stability automatic measurement of key dimension parameters of stud bolts.