Document extraction glitch

Trying to implement the technique described in this paper for extracting images from scanned documents: www.ee.columbia.edu/ln/dvmm/publications/99/spie-des.pdf

Done