Stereovision-based detection of 3-D weld seam using epipolar line constraint and laser stripe indication

Abstract

An epipolar line constraint equation is established for a binocolar stereovision system mounted at the end-effector of welding robot.The stereovision correspondence technique based on the combination of epipolar line constraint with laser stripe indication is applied to detect the position of a three-dimensional(3-D) saddle-shaped weld seam which is produced by the intersection of two circular pipes.The experimental results show that the smooth segments of laser stripe in the weld seam image can be obtained using thinning and deburring arithmetics,and the stereovision correspondences between pairs of points at the left and the right images can be dependably realized by detecting both the singular points of laser stripe curvature and the intersecting points of laser stripe with epipolar line,thus,the detection accuracy to the 3-D weld seam and its adjacent area can be improved.The geometrical shape of the 3-D weld seam is reconstructed from the 3-D data acquired by stereovision-based detection with less errors compared with its actual dimension,therefore the proposed method can satisfy the detection reqirment of 3-D weld seam in automatic robot welding system.