# Image processing for extracting regions of book pages that are covered by text

Is it possible to apply Mathematica's image processing capabilities for extracting regions of book pages that are covered by text, e.g. by a combination of blurring and binarizing, or better? Also the faster the better.

Here is a possible input and (approximate) output.

• @rasher Bluring, binarizing (as mentioned), and various filters mentioned under segmentation analysis. Unfortunately, the results are no good (not worth posting). – Drux Mar 13 '15 at 6:37

If you want a quick&dirty method, CornerFilter might be a good place to start:

img = Import["http://i.stack.imgur.com/HVRea.png"];
corners = CornerFilter[img, 2];
Closing[MorphologicalBinarize[corners, {0.2, 1}*10^-4],
DiskMatrix[2]]];