基于单视影像的地下通道3维重构方法

A Method of Subway 3D Reconstruction from Monocular Image

地下通道(包括隧道、巷道等)是一类重要的构筑物,重构其3维几何模型对于构建数字化操作平台有着重要的意义。在给定一个初始3维几何模型的条件下,利用地下通道影像上的线特征,通过人工交互方式获取相机的外方位元素及确定通道最终3维几何模型的参数,然后利用单视几何原理,通过反投影像点坐标即可计算得到详细准确的地下通道3维模型。实验表明该方法简捷有效,可以在实践中加以推广。

Subway （including tunnel, laneway） is a kind of important construction, and reconstruction of its 3D model from image is important for establishment of digital platform. Under the assumption that one initial 3D geometry model is given, camera extrinsic parameters, intrinsic parameters and parameters which determine the final 3D geometry model are obtained by interactive way using line features on subway image. Then using the principal of monocular vision theory, feature points on image is reprojected onto the 3D geometry model surface plane to get detailed 3D model. The experiment shows that this method is effective and can be developed in practice.