SULAIR SULAIR HomeSU Home
Robotic Book Scanning at Stanford University
A raw image from the DL before image processing. [Enlarge+]
 
 
Automated image processing

Automated image processing

When scanning is complete, the images produced by the DL are cropped and corrected for consistent lighting and contrast by BookRestorer image processing software, offered by I2S. BookRestorer also removes curvature in the book’s pages that results from scanning a bound object, and straightens pages to minimize skew of the text. Books scanned as a single image with facing pages are split into separate images, one for each page.

This completely automatic process results in a clean image for both left and right page. While effective, the process is not perfect. After it completes, an operator checks the images to ensure that all automated operations worked as intended, and that image quality is consistent throughout the book. Where automated image processing did not work correctly, the operator works directly with the original scan, and uses image processing tools specially designed for book restoration to produce a high quality image for the page.

At this stage, additional image treatment processes are applied on a project-by-project basis. For example, when digitizing art books, an ICC profile might be embedded in the images, or the book’s gamma adjusted.

Next »

©2004 The Board of Trustees of the Leland Stanford Junior University. All rights reserved.