| With the rapid development of the national heavy industry,the accuracy and speed of the straightness detection of rails and seamless steel tubes and the detection of deformation of H-beams after cutting are getting higher and higher in production.However,the detection of long and narrow components such as rail straightness,the degree of detection is still in the manual measurement stage,so this paper took the detection of narrow and long components as the research object,used machine vision detection to detect the long and narrow components,analyzed the factors that affect the detection accuracy of the long and narrow components,and proposed a high-precision detection algorithm.A complete set of detection scheme was proposed,and the straightness detection experiment of long and narrow members was completed,and then compared with the detection data of the laser tracker to verify the detection accuracy of the detection system.Firstly,this paper classified the research status of laser stripe centerline extraction algorithm,analyzed the reasons that affect the accuracy,and proposed an image segmentation method based on laser stripe color threshold,which reduced the influence of light to a certain extent.After the segmented laser stripe area was filtered by Gaussian filter,the sub-pixel of the center point of the stripe was extracted by polynomial fitting,and the degree of the polynomial was determined by comparing with the center point coordinate extracted by rough extraction.Finally,the algorithm was compared with the traditional extraction algorithm,and the extraction accuracy was verified.Based on the analysis of the factors that affect the accuracy of the existing calibration algorithm of line structured light plane,it was found that the lack of calibration points was the main factor that led to the low calibration accuracy.In view of this situation,the method of fitting the line structured light plane with space straight line instead of space point was adopted.Finally,the length of laser stripe on the standard gauge b lock was compared by the algorithm in this paper and the calibration algorithm based on the hidden line principle.The experimental results verified the accuracy of the proposed algorithm.In this paper,the existing point cloud splicing algorithm was analyzed,and a new point cloud coarse splicing algorithm was explored for this long and narrow component detection system.The common edge cloud coordinates of standard blocks were used as the reference points of point cloud splicing,and the average value of point cloud splicing matrix was calculated by taking the reference points for many times.Finally,the splicing matrix was iteratively spliced using ICP algorithm.Finally,aiming at the problem of low efficiency of error compensation in the detection process of long and narrow components,a compensation method for sloshing of long and narrow components in the detection process and lateral displacement of guide module was proposed by establishing external benchmark.Due to the large lateral deformation of the long and narrow components,considering the small field of view of the camera,the intelligent lateral guidance module was added.The change of the laser displacement sensor was transmitted to the single chip microcomputer to control the positive and negative rotation of the motor,so that the camera bridge could move back and forth,and realized the intelligent guidance of the camera bridge,so as to ensure that the tested parts were within the field of vision of the two cameras.Finally,the detection accuracy of the detection system was verified by comparing the detection data of the detection system with that of the laser tracker. |
