| Machine vision technology is a new technique of detection and recognition. It develops rapidly because of the development of all the relevant technology theory and the needs of actual production. And it has become increasingly broad prospects of the application in the national economy and national defense construction and many other areas.This article introduces the establishment of a set of visual sensing and image processing system based on the machine vision recognition technology as the core. The traditional detection technology of thread has some disadvantages like long processes, low efficiency and hard-line. It can achieve non-contact automatic detection and improve production efficiency to meet the actual demand of high efficiency of modern industrial production. The system uses the CCD camera to capture thread images, and it can automatically read the important geometrical parameters of thread by pattern recognition and digital image processing technology.This paper systematically describes the important theory of image processing and pattern recognition, and designs algorithms according to the characteristics of the thread contour. By collecting experiments, the original images of thread are obtained to be digital pre-processed, such as noise filtering, edge sharpening, threshold transforming. Secondly, the images need image segmentation of the partial images, including edge detection, contour extraction, refining and other processing. Then, we can extract the coordinates of thread corner points for the situation of tilt thread, fitted into a straight line, the thread will be horizontal screw rotation according to the slope of fitted straight. According to the characteristics of thread profile and the definition of the thread parameters, we can extract the teeth angle and teeth-type half-angle of the thread, find the large-diameter, path, pitch, pitch diameter and other important parameters of the thread. And finally we can identify the parameters of the thread types. |
PDF Full Text Request