(Date: 11.8.2014 – 14.8.2014)
I’m back from vacating, but will leave again Thursday afternoon.
During the time I’m at the institute a continue working on the hill climber and fix the rest of the bugs which are in.
I also look at another means of finding peaks: Something I call the
iterativeMaximumDeleter. Let me outline the idea: You have a bunch of hits belonging to a few tracks. The hits of one track will form small bands1 in the Hough space, which will overlap at one point. This point is equal to the track parameters. In a filled Hough space you now take the bin with the highest multiplicity, by means of some arbitrarily dumb maximum-finding routine. You delete all of the hits passing this highest bin from the Hough space. You start the procedure again with the new, highest bin. You stop until you either have an empty Hough space or don’t have any maxima any more.
For this to work I have to adapt my Hough Transform class to save origin information about every \((r, \alpha)\) pair. Some deep changes.
Sub bands, to be precise. I have to find some coherent terminology… ↩