We are showing some videos of our first results for the detections of flares based on the slit-jaw images (SJI) of IRIS.
We created a detection tool using some discrete cosine transforms - DCT complexity of O( Nlog(N) ) - in order to do fast analysis on our big data and to see them in the frequency domain.

As results of our detection tool, we produce videos showing the analysis and the output probability of flare on the global image, in blue, the probability is 0 and in yellow the probability is 1.

This image helps to understand how the frames are organised for the vizualisation of the results:

frames organisation for vizualisation

IMG: current image
Hist: histogram of the image
DCTa: analyse "after" the image
FPa: probability of flare obtained from DCTa
DCTb: analyse "before" the image
FPb: probability of flare obtained from DCTb


Here are some videos of our results:

Analyse for 20140329_140938_3860258481_SJI_1400
Partial flare (C class) at 0:08
Flare (X class) at 1:43

Analyse for 20140910_112825_3860259453_SJI_1400
Flare (X class) at 3:48


Loading Conversation