GIS Ostrava 2020




The role of Vienna‘s geoinfrastructure in Viennese Smart City Initiatives (Franz Xaver Pfaffenbichler, City of Vienna)

The workshop focuses on the Smart City initiatives of the City of Vienna and will point out the special role of the Viennese geoinfrastructure. Especially the leadership of the City of Vienna in publishing their geoadata as open data has a massive impact in new developments and innovations. Topics like „What means smart?“, „What means open?“ and „Can you rely on open?“ will be adressed as well as some concrete examples and best practices in our smart and open environment like, which is also accepted and appreciated above our borders. The presentation also gives an outlook on the Digital Agenda Vienna 2025, which deals among other things with the use of drones to measure areas and inspect facilities.

Towards real-time UAV mapping (Francesco Nex, University of Twente)

In this workshop an overview of the current trends in the development of UAV will be given showing how developments in adjacent domains are influencing the way unmanned platforms are used for Geomatics and Remote Sensing applications. Processing time is an important parameter in all the photogrammetric applications, but if fast computing is a plus, the use of UAV platforms for efficient monitoring and mapping have recently triggered the implementation of real-time solutions also in the Geomatics community. This workshop will present some early examples on how real-time or near real-time indoor and outdoor mapping applications can be implemented using commercial solutions too. More specifically, the real-time mapping of unknown indoor environments for quick inspections will be shown. The combination of SLAM, deep learning depth estimation and autonomous path planning algorithms allows to generate quick 3D models of indoor environments. On the other hand, two outdoor examples will demonstrate how real-time computing can be used for road monitoring or structural building damage detection after catastrophic events: SLAM and deep learning algorithms have been combined to detect objects of interest and geo-reference their position.

Advantages and disadvantages of the current implementations, future challenges for the development of a next generation UAV applications will be further discussed in the workshop. Participants will have insight on the implementation and will be able to understand how to produce similar solutions for their own applications.

Dense Image Matching and Multi View Stereo in Urban Areas (prof. Fabio Remondino, FBK Trento)

Image matching has a history of more than 50 years, with the first experiments performed with analogue procedures for cartographic and mapping purposes. The recent integration of computer vision algorithms and photogrammetric methods produced interesting procedures (also called Multi View Stereo – MVS) able to match every single pixel in corresponding images, deliver dense point clouds and thus automate the entire image-based 3D reconstruction process. Such procedures are fairly robust, but the final 3D point clouds often lack completeness and accuracy, especially while dealing with large complex scenarios, textureless surfaces, repeated patterns, obstacles or occluded areas.

The workshop will first review and analyze actual dense image matching algorithms (semi-global matching, patch-based stereo, etc.) and then run some experiments and comparisons with open-source solutions for the generation of dense point clouds. Geometric evaluations with ground truth data will be also performed.

Processing tools and datasets will be provided, participants must bring a mid-high powerful laptop in order to run the experiments.