[Issues] Satellite tracks not always detected

Hans Vaith hav at mpe.mpg.de
Mon Nov 13 19:04:06 CET 2006

Dear all,

when reducing data from the Sikkema dataset we found that satellite tracks 
were not always detected. Looking at the python wrapper code of the Hough 
transformation in SatelliteMap.py we did not find anything obvious, so I 
started looking into the C code of the hough transformation in the eclipse 
library. There seems to be a conceptual error or oversight. The 
transformation code does not account for the possibility of negative values 
for 'rho' (the perpendicular - or minimum - distance of the straight line 
from the origin of the coordinate system). rho is always negative when the 
offset of the straight line (b in the equation y = a*x + b) is negative.

I corrected the code locally on my laptop installation of AstroWise and ran it 
through the observations in question. All satellite tracks were then detected 
correctly (one track additionally needed tweaking of the HOUGH_THRESHOLD 
process parameter, because the path of the track through the image is very 

Our basis was CVS version 1.4 of hough_transform.c We could send a patch or 
our corrected source code, if there is interest. 


Hans Vaith   Max-Planck-Institut fuer extraterrestrische Physik
             Postfach 1312, D-85741 Garching, Germany
             Phone :(Country Code 49) 89 30000-3831
             e-mail: hav at mpe.mpg.de (Fax: 89/30000-3569)

More information about the Issues mailing list