The Research And Application Of Image Recognition In Laser Welding Locating System

Position measurement and Locating detection have a significant role in the area of manufacture and control of laser welding. Laser welding is a non-contact one, so it is quite difficult for the ordinary mechanical locating methods to meet the requirements in the process of accurate micro-laser welding.In the view of intelligent, automatic non-contact industrial processing and the development of manufacturing industry, the research objective and significance are put forward and one set of laser welding locating system based on machine vision is designed. And this thesis also emphasizes the application of image recognition in the locating system. Histogram equalization, median filter and edge detection are adopted to pre-handle the image, which achieves the goal of getting rid of noise, recognizing the edge and making the features of a target obvious.This paper proposes the detection of a straight line and a circle through HOUGH transforms on the basis of a pre-handling of images, the segmentation of images by using the optimal threshold value, the obtaining of the centre of a cirecle through the round fitting method of a circle. The extraction and calculation of image characteristic provide coordinate transformation parameters for positioning system.