[Image-SIG] How to divide a scanned image with multiple photos

[Philipp Bender]

Hi Franz,

that's not a trivial problem I think. You have e.g. first to create a mask 
with all pixels from 0..(sth like 20), this is the dark parts of the images 
plus the space around them (your scanner scans them as black). Then, you need 
to detect edges and apply a Hough transformation or something like that. 
Applying a Hough transformation is by the way very easy with openCV, maybe you 
can use openCV to get the edges and feed that to PIL to crop the interesting 
regions.

http://www.cognotics.com/opencv/docs/1.0/ref/opencvref_cv.htm

Scroll down to "Special Image Transformations" -> HoughLines2

Best regards
Philipp