This project was pursued under the direction of Professor Michael Gleicher in spring semester of 1999 by Craig R. Farrand. The goal of this project was to process scanned images of notes and produce quality output images to use for automated web creation, as well as regular viewing and printing.
In order to do this, we must accurately separate the background pixels from the ink pixels in the image. The ink pixels we want preserved in the final images are the definite ink pixels, as well as antialiased ink pixels that may or may not be in a local color space of the true ink color. The background pixels we want to remove are the pixels that compose the paper (lines, wrinkles, paper color, and scanner generated noise). For this project, removing the background consists of identifying background pixels and setting them to white. We must also resize the image to an 8 1/2 by 11 inch format which will make the image easier to view, as well as print.
The three major tools used to complete this project were Python, the Python Imaging Library (PIL), and SWIG (Simple Wrapper Interface Generator) which are available from http://www.python.org.