Research On Image Processing Of Laser Weld Seam Tracking Based On Morphological Feature Filtering

With the improvement of computer computing power,image processing technology is widely used in industry.In the field of automatic welding,linear laser vision sensing is considered to be the most effective sensing method.According to the linear laser projection method of linear laser vision sensor,it can be divided into direct type and oblique type.By comparing the accuracy of direct type and oblique type structure sensor,it is determined that the design of line laser vision sensor adopts oblique type structure,and gives the parameter selection standards for each component in the sensor when the sensor is assembled according to actual needs.In order to obtain the conversion relationship from the two-dimensional image coordinate information in the image collected by the sensor to the three-dimensional camera coordinate information,a sensor calibration method based on a checkerboard calibration board is designed in this paper,this method simplifies the steps of acquiring calibration images and eliminates human error factors in the calibration process.First,use Zhengyou Zhang's checkerboard calibration method to complete the calibration of the camera,and then complete the calibration of the structured light plane through the designed image processing method and the camera calibration result.As the core of visual sensing,Seam tracking technology has high requirements for the accuracy,real-time and robustness of image processing algorithms.In view of the redundant operations in the filtering process of most weld tracking image processing algorithms,the real-time performance is poor.In order to reduce the large number of redundant operations in the process of filtering the noise in the weld image,by analyzing the morphological characteristics of arc spatter and laser stripes in the weld image,and using the first-order Markov chain to describe the spatial correlation of the image matrix neighborhood,a laser weld tracking image processing algorithm based on morphological feature filtering is designed.When extracting the center line of the laser stripe,the algorithm removes the noise interference in the welding image by determining the morphological characteristics,effectively reducing the redundant operation of the filtering process,and improving the real-time performance of the algorithm.In order to verify the effectiveness of the image processing method proposed in this article,a series of offline tests are carried out using a data set covering typical working conditions.The results show that the filtering method based on morphological features can filter out the noise in the image,accurately extract the center line of the light strip,and has good accuracy,Robustness and real-time.At the same time,the actual welding test was carried out in this paper.The test platform is composed of a six-axis Fanuc welding robot and supporting equipment,a line laser vision sensor and an industrial computer.The welding seam tracking test is carried out for the lap welding type.The experimental results show that the real-time tracking welding can be completed with small tracking average error and smooth welding trajectory.