Abstract

A method for creating a 3D model of thin objects is disclosed. This method creates 3D models of each such thin object from cues taken from posed source images and 3D point clouds (which can come from depth maps of these images). The cues could be a vertically aligned line feature or discontinuity in a background image. The 3D models of the thin objects are then added to the input 3D point cloud, and a mesh reconstructed from this augmented point cloud. The above method allows easy creation of 3D imagery of near vertical or thin objects like poles or towers to be incorporated onto maps

Share

COinS