Visual Inspection And Path Generation Of The Four Dimensional Fillet Weld

In the development of modern intelligent industry,the four-dimensional fillet welds exist widely in containers,trucks,ships and other industrial equipment,however the coordinates and gaps of the four-dimensional fillet welds are difficult to be measured in different environments simultaneously.Thus,it's usually difficult to get accurate multi-dimensional parameters in industrial practice.In response to this problem,the visual parameters measurement of the four-dimensional fillet welds is proposed in this thesis.The system uses the four degree-of-freedom robot and four-dimensional fillet weld recognition algorithm to achieve the fillet weld path generation.The thesis includes the following parts:(1)The four-dimensional fillet weld visual recognition algorithm and intuitive visualization software architecture are designed.The visual sensor is employed to capture the weld seam image with laser stripe and send the digital image to the computer system.Then the image feature point can be extracted.(2)Four-dimensional fillet weld obvious feature recognition.Firstly,the Gaussian low-pass filtering is applied to smooth the edge of the fringe,and the brightness of the stripe center must be kept constant at the same time.The improved zhang fast parallel algorithm to extract the stripe centerline is proposed to filter out the tiny bifurcation on the laser stripe center line.Then the region of interest of the image is extracted by using the Hough transform to enhance the image processing speed and eliminate the interference of other arcs light.(3)Multi-exposure weak weld seam feature detection.When the four-dimensional fillet weld characteristics are weak,the four different exposure images can be collected in multiple exposure modes.Through the camera calibration curve and the known exposure time,we can merge the four images into one.Then,the thesis presents an adaptive growth algorithm to obtain the laser stripe centerline of the fusion image.The center line of the weld is extracted by the difference algorithm.The intersection of the two lines is the characteristic point of the fillet weld.(4)Four-dimensional fillet weld path generation is an important part of intelligent welding.The two-dimensional coordinates of the image are mapped to three-dimensional coordinates in the world coordinate system of the weld in the thesis.The industrial computer generated weld path so that the manufacturer can clearly see the weld seam path and the shape of the weldment and other information about the weld.The visual detection software not only visually shows the weldment area,but also intuitive displays weld seam trajectory in the interface.The experimental results show that the algorithm of weld seam feature point detection in this thesis that has the advantages of high precision,good stability and widely application in the field of welding technol.It can also extract the weld feature points accurately under the condition of arc light interference and weak weld characteristic.