Uni-Logo

Direktzugriff

    How a computer determines movements in videos

    25.09.2014 16:47
    As a PhD student 10 years back in time, the computer scientist Thomas Brox developed a seminal computer vision technique, for which he has now been awarded the renowned Koenderink Prize. Photo: Department of Computer Science, University of Freiburg

    Photo: Department of Computer Science, University of Freiburg

    How do you teach a computer to accurately determine movements in an image sequence?  Prof. Dr. Thomas Brox already addressed this problem as a PhD student. Now he has received the Koenderink Prize for a technique that he developed in 2004 together with his colleagues Andres Bruhn, Nils Papenberg, and Joachim Weickert. The award is one of the most prestigious prizes in computer vision. It recognizes works that have proved particularly valuable after one decade.

    The technique developed by Brox doubled the accuracy in motion estimation. Since then, many researchers have built upon this method and more than 1000 scientific publications referred to his work. Companies were also interested in the approach: motion estimation is required in modern driver assistance systems, which recognize dangerous situations and help avoid accidents.

    Thomas Brox is professor for pattern recognition and image processing at the department of computer science of the University of Freiburg and a member of the BIOSS Centre for Biological Signalling Studies since 2010. After finishing his PhD at the Saarland University in 2005, he worked at the University of Bonn, the Technical University Dresden and the University of California at Berkeley in the United States. His research focuses on video analysis and development of methods that enable computers to interpret the content of images.

    Vita


    Homepage

    Contact:
    Prof. Dr. Thomas Brox
    Computer Vision Group
    Department of Computer Science
    University of Freiburg
    Phone: +49 761 203-8261
    E-Mail: brox(at)informatik.uni-freiburg.de
     



    Benutzerspezifische Werkzeuge