Weld Appearance Inspection Based On Structured Light Vision

Welding is an important means of modern material processing.In actual processing,due to the influence of processing precision,the appearance of the weld often has various defects.The quality of welding directly affects the life and safety of welding products.The appearance inspection of the weld is an effective means to improve and ensure the quality of the weld.At present,the visual inspection of the weld is mainly carried out by visual measurement.The detection of weld defects mainly depends on the work experience of the staff.In addition,the visual inspection method is susceptible to environmental factors during the inspection process,so it is difficult to ensure the accuracy and efficiency of the test.Monocular vision has the advantages of non-contact,fast and real-time,and it is widely used in the fields of automatic drug sorting,robot path planning,target recognition,etc.It provides technical support for weld appearance inspection,In this paper,the fringe image of the weld profile obtained by structured light vision is used as the information source,and appearance detection algorithm based on structured light vision is explored.The algorithm overcomes the misjudgment and leakage caused by eye fatigue and empirical differences in manual measurement.The efficiency and accuracy of the welding appearance inspection are improved and the automation and intelligence requirements of the entire welding process are satisfied.The main research contents of this paper are as follows:In order to improve the quality of the acquired image,a set of structured light vision system for post-weld appearance inspection was designed.According the current research status and development trend,The detection principle of the structured light vision system was expounded,the software and hardware components of the weld appearance inspection system were determined,including the selection of cameras,lenses,filters and lasers.Based on the analysis of the mathematical model of the line structured light vision system,The Zhang template calibration method was used to calibrate the internal and external parameters of the camera.In the structured light plane calibration,only a simple two-dimensional planar target was used.In the hand-eye calibration,a two-step calibration algorithm was used.Experiments show that the calibration method of the detection system is simple,easy to implement and the calibration accuracy meets the requirements.Finally,the relationship between the image coordinate system and the three-dimensional system of the robot base coordinates was realized.This paper mainly studies V and I welding.Due to the existence of many noises in the structured light image,these noises will affect the treatment of the structured light.At this time,the weld contour stripe has a certain width,which affects the accurate extraction of the weld seam characteristic parameters,so the extraction of the center line of the structured light stripe is required.Due to the large interference factor in the acquisition process,the area filtering is needed to remove the interference of the isolated point,For a frame of image,we are only interested in the structured light stripe part of the image,so we choose the region we are interested in by column projection,a sub-pixel stripe centerline extraction algorithm based on hessian matrix method and gravity center method is adopted.In order to further improve the extraction precision of the weld centerline,the error points was removed and wire breakage compensation of the weld seam,finally the stripe centerline with better connectivity was obtained.Comparing the extraction algorithms of each centerline,the center line obtained by the method has high precision,low time consumption and high stability.It lays the foundation for the subsequent measurement of weld appearance inspection.In the process of measuring the dimensions of the weld,firstly,the dimensions of the weld bead are defined,in order to obtain the weld bead forming dimensions such as weld width and residual height,the feature points of the weld are extracted based on the traditional extraction algorithm.In this paper,an extraction algorithm combining recursive positioning and fitting method is proposed.According to the geometric characteristics of V and I weld,accurate measurement of weld bead size after welding was realized.In order to visualize the three-dimensional shape of the V-weld and the I-weld,the point cloud data is filtered,including discrete point removal,hole repair and point cloud down sampling.The Poisson surface reconstruction method is used to reconstruct the point cloud data to realize the weld seam visualization.In the three-dimensional appearance defect inspection of the weld,by comparing the morphology of the normal weld,the weld with the defects,the algorithm for detecting the surface defect of the weld is established,and the weld extracted by the method was obtained.The seam defect size information was measured by vernier caliper and the structure-based light vision system.The experimental results show that the weld surface defect detection method can accurately obtain the size information and position of the defect.That is,the weld defect detection algorithm based on the ideal contour is feasible.