The Research Based On The Double Line Structured Light Weld Feature Extraction

Since the welding automation concept presented, many scholars began to studyautomatic tracking technology. In the field of welding, recognized as the mosteffective way of sensor is the visual sensing method.Most domestic scholars study thesingle way of laser vision sensing, using dual laser vision sensing seam trackingresearch is less. In this paper, the dual structure light feature extraction are studied,dual laser sensor way compared to the single laser sensor, in reducing lead error,height recognition and prediction of welding seam track have more advantages.In this paper, designed of double seam tracking system including industrialcomputer, image acquisition card, biaxial robots and robot controller, double seamtracking sensors. Design of the double seam tracking adopt the way that two laserplaced vertically and the CCD camera received vertically. Sensor that using this wayhave the advantages of compact structure, small volume, easy to assemble. In thispaper, the camera that in the seam tracking system is calibrated, the cameracalibration precision is0.147mmAccording to the characteristics that the double line structured light sensor ishighly sensitive to height, the height of welding torch recognition model isestablished, and according to the installation of the relationship between weldingtorch and sensor,the welding torch deviation identification model is established. Thispaper proposes a anti-jamming strong image processing algorithms. In featureextraction, adopt the method of least squares linear regression to extract weld featurepoints.Through the experiment, the precision of the method is0.15mm, it take me10ms to process a frame image, can meet the requirements of welding.In VC++6.0development environment, with OpenCV1.0, author designed seamtracking software interface, the software interface mainly include the welding torch,image display, welding, long axis and short axis of the robot control, extraction ofdeviation display, etc. To test and verify the accuracy of image processing algorithm,designed in a given deviation were0.5mm,0.7mm,1mm and fixed Angle2.86and welding speed of7mm/s deviation extraction experiment, the experiment shows that the proposed algorithm is of high precision.