Bound for JavaPolis

On the 13th of december, Fabrizio Giudici and I will be giving a short talk (as a BOF session) on a Java-based parallel computing project of ours. This very application describes an application of parallel computing to image processing, with a view also to astronomy-oriented tasks.

You can find the full description of our BOF session at this link.

Many of you may know that parallel computing applications have recently come of age. The so-called “Moore’s Law” of computing power has recently started flattening due to the fact that the physical limits of the current microprocessor production process. It will probably be quite a long time before the next generation of quantum-based computers comes out: meanwhile, the only way to get more computing muscle from today’s technology is to integrate more microprocessor cores into one. Dual-core systems are very common now, and computers with even more processors will soon hit the mass-production market.

If you have no multicore computers, you can always try out distributed parallel computing, in which many single core units join together to perform very demanding tasks. The best-known such application is probably BOINC (formerly known as SETI@Home).

Some brainstorming Fabrizio and I had last spring led us to investigate new applications of grid computing to image processing. This ultimately gave birth to an open-source project of mine, Pleiades, which aims at providing a flexible toolset for image stacking and aligment, with quite a few perspective applications in digital photography and astronomical imaging. Pleiades in turn is largely based on Mistral, an open-source framework developed by Fabrizio targeted at providing a powerful, abstract imaging layer for image manipulation, which strongly supports parallel computing.

Finally, it is probably worth to say that both projects also make wide use of the Java Advanced Imaging (JAI) library.

Check back often for future updates!

Leave a comment