In the guiding system for fillet welding of corrugated sheet based on laser line projection, the Laser Lines Projected on the Corrugated Sheet and Plane (abbreviated as LLP_CS_P) are interfered by the irregular width of Assemble Gap to be Welded (abbreviated as AG-W), the Scratches of the Metal Surface (abbreviated as S_MSS), the Laser Line Projected on the Welding Point (abbreviated as LLP_WP) an so on. As a solution of these problems, an algorithm and on-line processing system based on morphology are proposed to stably and accurately locate the endpoints of LLP_CS_P. First, the interferences are separated from the laser line projection. Second, the interferences are eliminated by morphology filter. Third, the locations of laser line projection at AG-W are obtained through connected component labeling and extreme theory. Image acquiring and processing on line indicate that this algorithm and system can perform stably and effectively under the condition of irregular width of fillet welding seam of corrugated sheet, scratches, welding points. The trajectory and width of welding seam are monitored on line to guide the torch to track the AG-W and adjust welding parameters adaptively.
Orientation detection of fruits by means of convolutional neural networks and laser line projection for the automation of fruit packing systems