The imagery sample contained a mixture of large commercial buildings and private houses© DigitalGlobe, copyright 2011. All rights reserved
JRC scientists win a Data Fusion Contest
A team of JRC scientists were among the winners of the Data Fusion Contest 2011 which challenged scientists to use the same set of high resolution spatial imagery to show their technical advances. The two other winning teams were from the Rochester Institute of Technology and the University of Montreal. The competition was organised by the Institute of Electrical and Electronics Engineers (IEEE), a non-profit professional association dedicated to technological innovation in the electrical engineering, computer science and electronics. It counts over 400 000 members in some 160 countries, 45% outside the United States.
This year contest was focused on multi-angular acquisitions over the same target area. Given the wide choice of possible applications, participants were free to choose the research topic. DigitalGlobe, a US commercial vendor of space imagery and geospatial content, provided a set of multi sequence images from its satellite WorldView 2. The high.resolution imagery, collected over Rio de Janeiro, Brazil, included a mixture of large commercial buildings and private houses, as well as the airport and some community parks.
The team from the JRC Institute for the Protection and Security of the Citizen (IPSC) presented a novel method for measuring the height of objects using satellite images, without the need of satellite positioning. Results focused on building heights but the method can be applied to any object in the scene. The method was applied to the full scenes and validated against known building heights acquired from public sources and manually calculated with accuracies of less than 1 m for building heights ranging between 8 and 160 m.
The JRC method was presented under the title ‘Building height extraction from multi-angular WorldView-2 Ortho Ready images using GPU-accelerated template matching’.