The tremendous increase of computer power facilitates tackling more and more complex problems based on more and more refined method, thus representing real life processes better and better. A steadily growing wealth of data (taking often the form of a large image or of a video stream) is thereby becoming available and it becomes harder and harder to store, access, evaluate and further process the corresponding information. Areas like multimedia, telecommunications and imaging sciences are evident examples of fields where efficient coding and processing techniques become essential.
In this framework the image processing tools need to couple high effectiveness (i.e. high capacity of profiling and correctly identifying patterns of interest, which can be of a higly different nature, depending on the application at hand), with high processing speed (many applications demand that the analysis be carried out in real time). All this should be achieved at the lowest possible price, so that the tools can be easily available to potential users.
The goal of the project is to systematically study the mathematical foundations and the algorithmical aspects of the different technologies available for the design of image processing tools in view of application in the fields of medical imaging and of biometrics. Our aim is to compare different approaches in order on one hand to improve the existing methods, and on the other hand to facilitate, or, in the long term, to automatically adapt the choice of the right tool to the particular scenario at hand.