Mobile Mapping and Visualization of Indoor Structures to Simplify Scene Understanding and Location Awareness

Abstract

We present a technology to capture, reconstruct and explore multi-room indoor structures from panorama images generated with the aid of commodity mobile devices. Our approach is motivated by the need for fast and effective systems to simplify indoor data acquisition, as required in many real-world cases where mapping the structure is more important than capturing 3D details, such as the design of smart houses. We combine and extend state-of-the-art results to obtain indoor models scaled to their real-world metric dimension, making them available for online exploration. Moreover, since our target is to assist end-users not necessarily skilled in virtual reality and 3D objects interaction, we introduce a client-server image-based navigation system, exploiting this simplified indoor structure to support a low-degree-of-freedom user interface. We tested our approach in several indoor environments and carried out a preliminary user study to assess the usability of the system by people without a specific technical background.

Fabio Ganovelli
Fabio Ganovelli
Senior Researcher

Senior Researcher at the Visual Computing Lab