Weld Seam Tracking and Detection Method Based on Structured Light Vision

Tracking and detecting weld seams are essential tasks in weld seam inspection. Traditional weld seam detection methods primarily rely on manual measurement using vernier calipers, which suffer from poor adaptability and low sensitivity. In this work, we introduce a tracking and detection approach for weld seams that is founded on structured light vision. The process begins with a projection device casting a multi-line structured light pattern onto the weld seam's surface. The camera captures the reflected image and sends it to the control unit. Following this, the three-dimensional (3D) coordinates of the weld feature points are determined through a sequence of camera calibration, structured light calibration, and feature extraction. This capability enables both dynamic tracking and a full 3D reconstruction of the weld seam. Validation experiments confirm that our proposed method delivers exceptional performance, characterized by high precision, robust accuracy, and reliable real-time operation for both tracking and detection tasks.