Multimodal fusion-based system for wall thickness measurement of flexible dust cover

As key vibration-damping protective parts, automotive drive shaft dust covers rely on uniform wall thickness to guarantee drivetrain reliability. Conventional structured light methods exhibit inherent limitations on black low-reflective corrugated workpieces: low fringe contrast and edge phase jumps blur or erase point cloud edges, while noise-induced false extrema undermine mass measurement reliability. To address drawbacks including edge detail loss, insufficient measurement accuracy and poor industrial adaptability, this paper proposes a wall thickness detection method using structured light 3D measurement, fusing 2D images and 3D point clouds. Firstly, the Scharr operator extracts 2D image edges, incorporated into the 3D reconstruction process as constraints for accurate pixel-point cloud matching. Subsequently, 3D point cloud edges are extracted using local geometric features. Finally, 3D edge point clouds are orthographically projected onto the 2D plane, combined with point cloud curve extremum detection using local smoothing and adaptive window NMS and cross-curve distance calculation, achieving accurate wall thickness measurement. Experimental results reveal that the proposed method restricts the overall error rate below 10% and achieves measurement precision within 0.1 mm. Compared with conventional schemes, it raises detection accuracy by an average of 9.39% and supports dust covers of diverse specifications.