narrow-cut-up

Hypothesis

We hypothesize that we can do better than naive-cut-up.

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

• Like naive-cut-up, except that it:
  • scales each image to the size of the base ("canvas") image
  • cuts and pastes in narrow strips instead of 1/9-page chunks

Usage

In this example, we use images already obtained used by naive-cut-up fetch-chronam and sorted as described in

$ ./narrow-cut-up.py ca_5.jp2.png pages/*

There are also some command-line options which are not yet documented.

Observations

The result has promise, but looks like it might work best with pages which contain more text than images.