Here are my notes on the picky details of the math and
code involved in the color2gray optimizations.
I've also put together a simple commandline
(Public svn is at:
compiled the code in gcc, as well as msvc7, and it ought to be fairly portable.
Both the code and the notes should be considered "rough preliminary versions",
if you find any typos etc, please drop me an email. The results that I'm
getting from the neighborhood cases, while they seem fairly sensible, are a
little different than the results that we got using the original OpenCV code --
I'm not sure why this is, or when we may get around to tracking down whatever's