Objective In response to the common segment staggering disease in intercity railway shield tunnels, there is an urgent need for an automated detection method that combines high precision, lightweight, and low cost to enhance detection efficiency and accuracy.

Method High-density point cloud data of the segment ring joint area are acquired using a structured light binocular vision camera, and then preprocessed through methods such as filtering, correction, and down-sampling. To address the ring joint detection problem, the point cloud is converted into a grayscale intensity map. Initial edges are extracted by combining Canny edge detection and morphological processing. The Hough transform is employed to detect straight lines, and a dual-threshold clustering algorithm incorporating direction and distance information is introduced to achieve precise identification of ring joint lines. For staggering detection, the target measurement area is automatically annotated based on a region-growing and KD(K-Dimensional)-tree optimization strategy. The feasibility and accuracy of the equipment are verified by comparing the interactive staggering extraction method with manual measurement results. Two staggering extraction methods are proposed, i.e. a partitioned plane fitting method based on local elevation differences and a global quadric surface fitting method. An image stitching framework integrating SIFT (scale-invariant feature transform) feature point extraction and RANSAC (random sample consensus) robust registration is constructed to achieve seamless fusion and panoramic stitching display of the staggered segment images.

Result & Conclusion Test results from shield tunnels in the intercity railways of the Guangdong-Hong Kong-Macao Greater Bay Area demonstrate that the structured light binocular vision camera can achieve an accuracy of ±0.2 mm in staggering detection, providing a new technical approach for the full lifecycle monitoring of intercity railway shield tunnels.