Feature point extraction algorithm of thick plate weld based on line structured light
Aiming at the difficulty in feature points extracting of thick plate structural parts using the laser vision system,caused by difficulty in image feature region extracting and the uneven diffuse reflection of workpiece surface processing,a feature point extraction algorithm for thick plate welds based on line structured light is proposed.Firstly,the laser vision system obtains the weld image and uses the YOLOv4 algorithm for pre-training,and uses the weight file obtained by training to automatically detect and obtain the weld feature region of interest(ROI).Secondly,the ROI is processed by noise reduction,binarization,etc,and the center line of the weld is obtained by line by line or column by column search method.Finally,according to different weld types,the distance method and straigth line segment clustering method are used to extract feature points based on the least square method.The experimental results show that the method can effectively extract different types of weld feature points,and has the characteristics of strong robustness and small identification error.
NETL
NSTL
万方数据
维普
