Walker da Silva, C., Rosa, Reinaldo R., Costa, J. E. R.

The study of the magnetic fields present in the solar corona is attracting increasing interest due to strong evidence of its relation to the phenomena observed in that region of the solar atmosphere. From the last two decades we have available images of the solar corona with increasingly better resolution, so that the comparison of these images with the physical models developed to study the dynamics of coronal magnetic fields is a task of extreme importance in solar astrophysics. We present in this work a new approach for the use of Hough Transform (HT) in order to extract features from the solar coronal images obtained in the Extreme Ultra-Violet (EUV). Our methodology is based on the fragmentation of the image and using of Hough Transform to get results from each fragment, and then identifying the fragments of the same magnetic field line using an interconnection algorithm. Finally we developed the evaluation of our methodology, where we proceed to compare the pixel coordinates of extracted magnetic field lines with those generated synthetically by the physical model. The features extraction algorithm is still under evaluation, but the early results have proved satisfactory.

hough transform, features extraction, pattern recognition, computer vision, solar astrophysics.