Indoor 3D measurement using camera pose estimation based on line features

Abstract

This study proposes a measurement method for the interior of large structures for infrastructure maintenance and inspection. The 3D point cloud of a room is obtained by a light section method using sphere camera that can capture a 360-degree view of the surrounding area and a 360-degree ring laser. Specifically, the camera's orientation is determined by detecting line components in real space using the spherical camera images and comparing the distribution of these line components with the three directions in real space, based on the Manhattan world assumption. Using the camera's orientation, the 3D point cloud of the entire room is generated by integrating the point clouds obtained from multiple light sections.