We have made use of computer vision techniques to clean
the
USNO-B Catalog of spurious entries that have been caused by
diffraction spikes and reflection halos in the original imaging. The
paper describing this work is available as
Barron et al (2008).
The table of spurious sources found by this project can be found
here.
A tarball of the code developed for this project can be found
here.
The released data is a FITS file containing a "diff" of all of USNO-B. It is a list of the spurious sources we have found in USNO-B, not a new version of USNO-B with the spurious sources removed. Therefore, to use this data, one must first be in possession of the USNO-B 1.0 Catalog, and one must manually remove those entries in the catalog that are present in our "diff" file. Details on the conditions of use of the released data, regarding credit and academic citation, can be found here, and also within the data tar file.
The code is intended only to validate the algorithm as it was explained in Barron et al 2008, and is not intended to be executed. It is all "research" code, and as such it is often inelegant and poorly commented. There is a README file in the tarball that contains a short explanation of the structure of the code, and each file contains brief explanatory comments.
Questions regarding this project should be addressed to .