I was wondering if y'all would be interested in running some of the winning entries against a new set of images once the contest is over. I fear that over the past few days, the solution has been mostly tailored to the problem and cannot yield any reasonable good score if given new inputs. Entries like this one ( http://www.mathworks.com/matlabcentral/contest/contests/2/submissions/4338 ) on the other hand is likely to produce better result with a different set even though it is now far away in the ranking system.
BTW, I write a small blog on compressed sensing and some readers and I have discussed some of the issues arising from this very interesting contest.
http://nuit-blanche.blogspot.com/2010/05/cs-matlab-programming-contest-on_04.html
http://nuit-blanche.blogspot.com/2010/05/cs-one-fallout-of-matlab-programming.html
If proven wrong on the fact that some of the competing solutions are in fact doing nominally better than the entry just mentioned then I think it would be an extraordinary good news to hear about it and it ought be part of a paper.
Igor.