Research On Weld Image Processing Based On OpenCV

With the continuous development of the social economy,it has also achieved rapid development in the fields of modern industrial manufacturing such as automobiles,machinery and aviation.Welding is a very important part of these industrial manufacturing processes.Welding technology is also one of the symbols of the national machinery manufacturing and scientific and technological development level.In order to improve productivity and production quality,welding robots began to replace manual welding.However,most of the welding robots at this time are teaching reproduction robots,which require a relatively high welding environment.During the welding process,the workpiece will be deformed by heat.Since the trial teaching reproduction robot only performs welding according to the established welding trajectory,it is impossible to accurately identify the position of the welded seam after the deformation,thereby causing welding quality problems.In order to overcome this defect,the weld seam identification tracking system came into being,which can track the weld seam in real time,guiding the robot to move on a precise trajectory for intelligent welding.This paper refers to the research status and related technology of welding robot weld identification at home and abroad.For a narrow V-shaped weld with a width and depth of less than 1cm,a set of weld identification system based on OpenCV software is designed.The system includes a word line laser and a CMOS camera.It uses direct laser triangulation.The CMOS camera captures images in real time and transmits them to the host computer.Obtain the feature points of the weld by processing with the OpenCV algorithm in the host computer.Then we can calculate the displacement of the weld in both vertical and horizontal directions by the position information of the feature points in the two images.Then,these data can guide the welding robot to accurately weld andrealize intelligence.The core part of the system is an image processing algorithm based on OpenCV programming.In the pre-processing stage,the image is filtered by mean filtering,median filtering,and Gaussian filtering,respectively,and Gaussian filtering is finally selected according to the actual effect.Through theoretical analysis,the welding image is binarized using the Otsu algorithm with an automatic acquisition threshold.Then,by comparing the actual effects of Robert,Sobel,Prewitt,LOG and Canny edge detection algorithms,the Canny edge detection algorithm is selected.After the edges are acquired,the upper and lower pixels are scanned to obtain the center line.Finally,the Harris corner detection algorithm is used to obtain the weld feature points.The OpenCV-based weld seam identification system designed in this paper has a high degree of automation and is very promising.