Vision based mobile mapping

This talk briefly introduces the research group on 3D Geoinformation Technologies of the FHNW University of Applied Sciences and Art Northwestern Switzerland and presents some of the research highlights, including virtual 3d globes and 3d cultural heritage. It then focusses on ongoing activities in 3d vision based mobile mapping which recently led to the founding of a new spin-off company (www.inovitas.ch).

The research on vision-based mobile mapping relies on a multi-stereovision approach covering the entire process chain from the acquisition to the cloud- and web-based exploitation. The acquisition system currently features GNSS/INS positioning, multiple stereo camera systems capable of capturing high-resolution stereo image sequences with data volumes of > 1 TB per hour. The technology features an absolute 3d accuracy potential of 3-4 cm and a relative measuring accuracy of better than 1 cm. The processing and exploitation pipeline for the Multi-TB stereo imagery data sets includes geometric and radiometric processing of the imagery as well as depth map generation based on dense stereo matching. It processing pipeline was designed for high-performance parallel processing in High-Performance Computing (HPC) and Cloud-based Computing environments. For the web-based exploitation we are currently developing a stereovision mobile mapping SDK based on WebGL and HTML5, thus exploiting numerous synergies with the Institute's OpenWebGlobe activities. The SDK allows the integration of monoscopic and stereoscopic viewing as well as accurate 3d measuring functionality into different Web-GIS environments.