Skip to main content

Integration of concepts for the parallelization of image processing algorithms into parallel compiler technology

Objective



The main focus of the project is to include concepts for the explicit parallelization of algorithms from the area of image processing and pattern recognition into modern parallelizing compiler and run time system technology. Massively parallel image processing systems require a very high bandwidth and have to exploit the communication facilities of the parallel system as far as possible. Application adapted data distribution schemes will be investigated for several algorithmic classes. Those include static and data dependend algorithms, e.g. scale space and multigrid methods. Based on the already developed concepts for explicit parallelization methods and the research on new algorithmic classes the results will be included in the Vienna Fortran Compiling Systems. The work will form a basis for general purpose parallel image processing software, that is efficient and easy to implement. Major stimulations for new applications in the area of image processing and parallel compiler systems are expected.

Funding Scheme

RGI - Research grants (individual fellowships)

Coordinator

Universität Wien
Address
22,Liechtensteinstraße 22
1090 Wien
Austria

Participants (1)

Not available
Germany