Feature detection in video

I have really had some fun with this program. It is a 5 funny on a LOLcats scale, where 1 is 10 LOLcat videos, 2 is 100 videos, and  so a 5 is 10⁵ LOLcats. For the layperson that is a "Category Five" or "CAT 5" LOL, which is not at all like a CAT 5 ROFL which is lethal.
 The video referenced here is an MIT lecture on optics, IIRC. Isn't that like product placement on TV? I suppose it can't be helped without painting over the trademarks on everything.
It is C++ code using opencv on Linux doing "Feature Matching" code at opencv.org. It is specifically the FLANN based matcher.

Just for reference, it is compiled to a.out like:

g++ SURF_FlannMatcher.cpp `pkg-config --libs opencv`

Then run with two images, the first being the desired thing to find and the second is the image to find it in.

 Bash Command:./a.out mathlf.png mathl.png

My PS1 is different as I like to know when I am in Python or octave or some other shell quickly so I set

PS1="Bash Command:"