Exploring terrestrial point clouds with Google Street View for discovery and fine-grained catalog of urban objects.

Localization and annotation of objects of interest can become an exhaustive task in dense point clouds without color information. Previous studies used point cloud data with aligned optical images for better scene understanding, however, are limited by huge data acquisition and system setup. Geospatial technologies are increasingly relevant to automatic land survey, but their use may be limited by cost and domain expertise. Technologies like Google Street View are appealing because they are unconfined and easy to operate on a web browser. We proposed a module to interconnect Google Street View client and The Visualization Toolkit (VTK) for terrestrial point cloud exploration. We evaluate the proposed pipeline with large-scale terrestrial laser scanned point-clouds with over a billion points in complex outdoor scenes. The integration of Google Street View and point cloud data provide better scene understanding and efficiency point cloud annotation than annotating point cloud without color information. Our results demonstrate that interactive modules using Street View may be suitable for self-referencing a surveyor within the point cloud and assist survey tasks more efficiently than software-based and field surveys.