UNLocX - Uncertainty principles versus localization properties, function systems for efficient coding schemes

Logo UNLocX

Researcher: Jan Hendrik Kobarg, Sabine Eifeld, Peter Maaß
Project funding: Europäische Union
Project sponsor: Europäische Kommission (EC)
Partner: Pierre Vandergheynst, EPFL, Schweiz; Bruno Torrésani, Université de Provence, Frankreich; Nir Sochen, Tel Aviv University, Israel; Hans-Georg Stark, Hochschule Aschaffenburg; Hans Georg Feichtinger, Universität Wien, Österreich; SagivTech Ltd., Israel; Steinbeis Innovationszentrum SCiLS, Steinbeis Innovation gGmbH, Bremen; Genesis S.A.; European Research Services GmbH, Münster
Duration: 01.09.2010 - 31.08.2013

The topic of the project is in the field of modern information and communication technologies: The volume of data here continues to rise steadily. Already today, audio and video applications demand ever higher computer performance and produce more and more data traffic. Although new developments are also continuously improving processor performance, the demand for modern software applications is increasing faster than the available resources. Developments such as the now established MP3 compression for audio data or MPEG compression for video data already take this demand into account to some extent.

With UNLocX, a similar approach is now being pursued and a new generation of signal processing algorithms is to be tested, which - for example in medical image processing - enable the processing of problems whose complexity was previously too high and, in addition, enable even more efficient compression and data transmission. The focus is on the development of the mathematical foundations and the implementation of a software framework, a developer platform, for highly efficient signal coding, transmission and processing, especially of audio and video data. The mathematical basis for this is provided by function systems that have optimal localisation properties and can be discretised efficiently. "It has long been known that nothing is given for free in signal processing: for example, if you wanted to measure the pitch of a tone very precisely, this tone would have to last indefinitely. Such 'fuzziness principles' exist everywhere in physics and signal processing, and the aim is to develop algorithms that - to put it simply - make the best of them," is how the project's approach can be described.

The practical development and testing of the UNLocX systems takes place in close coordination with the participating company partners on potential applications. One example is mass spectrometry in drug development. While it was previously only possible to carry out associated evaluations at selected points of two-dimensional tissue sections, the development of new algorithms could contribute to the development of genuine three-dimensional analyses and associated visualisations of results.