Automatic Segmentation of Fuzzy Laser Lines with Sub-Pixel Accuracy from the Uneven Background During Robotic Arc Welding

Robust and automatic segmentation of the reflected laser lines from the arc light modified background is prerequisite for the subsequent measurement of the weld pool shape which is of great importance for monitoring robotic arc welding. In this paper, the two dimensional intensity distribution caused by the arc light in the captured image is modeled. Based on the model, we propose an efficient and effective method that comprises three parts: (1), spline enhancement to achieve sub-pixel accuracy, (2), a gradient detection filter to eliminate the uneven background, (3), an effective threshold selection method. Experimental results verified that the proposed method is significantly superior to other state of art methods in segmentation accuracy.