A method and system for nondestructive testing of data line appearance defects based on machine vision

Abstract

The application discloses a kind of based on machine vision's data line appearance defect nondestructive testing method and system, belong to machine vision detection technical field, obtain the image sequence of data line in the image sequence of the data line to be detected in conveying line, and extract edge contour information, obtain contour feature set;According to the curvature distribution parameter of transition area calculated from contour feature set, and based on curvature change trend determines suspected topographic anomaly region set;Suspected topographic anomaly region extracts gray gradient feature and surface texture feature, constructs the standard appearance model of data line transition area, and calculates the topographic deviation value between suspected topographic anomaly region and standard appearance model;According to topographic deviation value, defect region set is obtained by defect determination, whether data line exists appearance defect is determined according to defect region set and output detection result;The application can realize the accurate identification of the tiny topographic anomaly of data line transition area, improve the accuracy and stability of appearance defect detection.

Images