Aiming at the problem of low working efficiency and difficult to meet mass

online detection requirements of traditional external thread detection method

an efficient

precise and non-contact detection method based on corner detection algorithm is proposed to detect the parameters of external thread by machine vision technology

digital image processing technology and machine learning. For the purpose of controlling thread processing quality

use CCD camera to acquire thread image

after image processing

extract the effective information of contour function and corner point coordinate

and point set classification operation is carried out by using support vector machine（ SVM）. Through data analysis

the parameters of range of thread

middle diameter

large diameter

small diameter and thread angle can be measured

achieve its aim of controlling thread processing quality. Test results show that parameters detection of 200 screw thread parts every minute can be completed

detecting speed is up to 3. 3 pieces/s

the detection precision is up to 0. 001 mm.