Mesh-based 3D Textured Urban Mapping

摘要

In the era of autonomous driving, urban mapping represents a core step to letvehicles interact with the urban context. Successful mapping algorithms havebeen proposed in the last decade building the map leveraging on data from asingle sensor. The focus of the system presented in this paper is twofold: thejoint estimation of a 3D map from lidar data and images, based on a 3D mesh,and its texturing. Indeed, even if most surveying vehicles for mapping areendowed by cameras and lidar, existing mapping algorithms usually rely oneither images or lidar data; moreover both image-based and lidar-based systemsoften represent the map as a point cloud, while a continuous textured meshrepresentation would be useful for visualization and navigation purposes. Inthe proposed framework, we join the accuracy of the 3D lidar data, and thedense information and appearance carried by the images, in estimating avisibility consistent map upon the lidar measurements, and refining itphotometrically through the acquired images. We evaluate the proposed frameworkagainst the KITTI dataset and we show the performance improvement with respectto two state of the art urban mapping algorithms, and two widely used surfacereconstruction algorithms in Computer Graphics.