中文摘要：

提取隧道横断面通常是基于隧道中线，而隧道中线提取是最困难的一个环节．本文提出了一种有效的隧道全局中线自动提取算法：通过从隧道的点云水平投影中提取边界点，并利用随机采样一致性算法提取边界点中用于多模型拟合的内点，最终的隧道中线基于再次随机采样的全局优化算法确定．所提出的算法被应用于位于四川的写字岩隧道，结果表明：全局中线提取达到了较高的精度，其均方根误差为16．5mm；与全站仪测量的隧道中线点相比，偏差的均方根误差为10．3mm．

英文摘要：

Terrestrial laser scanning has been widely used in tunnel engineering, the extraction of tunnel cross sections is often based on tunnel centerlines, which is the most difficult part.A novel method for automated extraction of tunnel centerlines was presented in this paper.In this process, the boundary points of a tunnel are extracted from tunnel point clouds,the inliers for multiple model fitting are estimated from tunnel boundary points using RANSAC （RANdomSAmple Consensus） algorithm,the final centerlines are determined using a global optimization based on a re-sampled algorithm.The proposed method was applied in the Xieziyan tunnel in Sichuan. The results of application showed the extraction of tunnel centerlines achieved a high accuracy.The RSME of centerlines fitting was 16.5mm.Compared with total station surveying,the RMSE of the deviations was 10.3 mm.