Machine recognition of laser reflection from gas metal arc weld pool surfaces

The reflection of projected laser lines may be used to determine the three-dimensional geometry of the reflecting weld pool surface. However, for gas metal arc welding (GMAW), the transfer of the droplets into the weld pool makes the weld pool surface highly dynamic and fluctuating. The position and geometry of the local reflecting surface, which intercepts and reflects the projected laser changes rapidly. As a result, the reflection rays change their trajectories rapidly. The contrast of laser reflection with the background is much reduced and methods are needed to extract laser reflection from low contrast images. To this end, an image quality measurement method is proposed based on the number of the edge points to determine if an image may be further processed. The image to be processed is then modeled as a superposition of the laser reflection and arc radiation background. Methods have been proposed to remove the uneven distribution of the arc radiation background from the image, such that a global threshold is possible to segment the laser reflection lines. The set of the laser line points are then clustered to form separate laser lines. These laser lines are then modeled and the parameters in the models are used to validate each modeled line. Processing results verified the effectiveness of the proposed methodsalgorithms in providing laser lines from low contrast images that are formed by laser reflection from a high dynamic gas metal arc weld pool surface.