Detection and Tracking of Coronal Mass Ejections (CMEs) by Means of the Watershed Segmentation and Hough Transform

Detection and tracking of CMEs is an important issue in solar physics. The appropriate images are continuously collected and parameters describing the extent, position, and time of occurrence of the CMEs are put in a catalogue. Because of the huge amount of data, it is imperative to process these images automatically. One of the best techniques for object extraction is the watershed segmentation. It works well for individual images. It is shown in this paper how this technique can be extended to a sequence of images showing the temporal development of the CMEs. The detection and tracking of the CME is done automatically via the Hough transform which points to at least some pixels located on the front of the moving CME. An approach has been developed which allows one to obtain the initial masks of the CMEs and subsequently merge the regions obtained from the watershed segmentation into the full masks of the CMEs so that in response to the sequence of CME input images a sequence of masks is generated.