find-cut-up-regions

Hypothesis

We hypothesize that it is possible to use Pillow to find "sensible" regions of an image to use in a cut-up construction.

Apparatus

• Python 2.7.6 (probably works with older versions too)
• Pillow (it might work with PIL too)
• Some scanned images of newspapers, books, etc.

Method

• First, decide what makes a pixel "light" versus "dark".
• Crop (say 5%) off the edges of the image to account for scanner darkness.
• For each horizontal line of pixels in the image, if some percentage (say 90%) of the pixels are "light", record this line.
• Amalgamate adjacent recorded lines and possibly reduce their thickness.
• Strips to cut will be those between the lines.
• Then, for each strip, rotate it 90 degrees and run in through the above process (as if it were an entire input image), rotating the generated "strips" from that back -90 degrees.

Observations

So far, so good.