April 12, 2013, 12:00 pm - 1:00 pm

April 12, 12:00 pm - 1:00 pm

Computer Vision for Solar Physics

Petrus C Martens, Physics and Computer Science Departments, Montana State University, Bozeman, MT, USA and Smithsonian Astrophysical Observatory, Cambridge, MA, USA

The Solar Dynamics Observatory (SDO) data repository will dwarf the archives of all previous solar physics missions put together. NASA recognized early on that the traditional methods of analyzing the data -- solar scientists and grad students analyzing the images by hand -- would simply not suffice, and tasked our Feature Finding Team (FFT) with developing automated feature recognition modules for solar events and phenomena likely to be observed by SDO. The purpose of these modules is to produce metadata (data about the data, like catalogs) that enable solar scientist to conduct statistical studies involving large sets of events that would be impossible now with traditional means.

We have followed a two-track approach in our project: developing task-specific solar feature finding modules to be "pipe-line" ready for the stream of SDO data, and developing an entirely new "trainable" module capable of identifying all types of solar phenomena starting from a limited number of user-provided examples. Both approaches have now reached fruition.

My presentation will consist of three parts. In the first parts I will report on the status and results of our 15 task specific modules, showing examples through movies and imagery. In the second part of my presentation I will focus on our “trainable” module, which is the most innovative in character: our approach represents the beginning of a more human-like procedure for computer image recognition.

In the final part of my seminar I will present examples of the data-driven solar science that is enabled by automated feature recognition.