Pablo Ruiz, Olivia Dill, Goutam Raju, Oliver Cossairt, Marc Walton and Aggelos K. Katsaggelos
in Digital Applications in Archeology and Cultural Heritage, vol. 15, December 2019, e00121.
DOI:10.1016/j.daach.2019.e00121
PROJECT DESCRIPTION
Historical paper often contains features embedded in its structure that are invisible under standard viewing conditions. These features, in particular watermarks, laid lines, and chain lines, can provide valuable information about a sheet’s provenance. While many methods exist for visualizing and documenting paper structure, each method must strike a balance between precision of reproduction, expense, and ease of operation. Standard methods of reproducing watermarks automatically, such as beta-radiography and low-voltage x-rays, are costly, time and training intensive, and therefore inaccessible to many institutions or individuals. In this work we propose as an alternative method the use of visible light and consumer-grade equipment to document watermarks. We introduce an image acquisition procedure, image processing suite, and inexpensive prototype whose main elements are a light table and a consumer-grade, digital, photographic camera. For a given document we acquire an image with light emitted by a light table passing through the document and a second image of the front side with ambient light. We additionally acquire a third ambient-light image of the back side for double-sided prints. The images are then registered to suppress the printed elements and isolate the watermark, using an algorithm also described here. The proposed method is capable of recovering images of watermarks similar to the ones obtained with standard methods while being a non destructive, rapid, easy to operate, and inexpensive method.
ACQUISITION SYSTEM
PROPOSED METHOD
The camera is mounted on a tripod and leveled perpendicular to a light table on a flat surface below (see Fig.1). Reflected light images were taken in ambient room lighting provided by overhead lights. The document is placed on top of the light table and a black cardboard frame is placed on top and weighted down. The frame secures the document in place, blocks light from parts of the light table that are not covered by the document, and landmarks the position of the paper within the camera’s field of view. With the recto facing upward, one image is captured with only room lights illuminated and a second with only the light table illuminated, named transmission image. In the case of documents with ink on both sides, the page is then flipped and realigned with landmarks on the black cardboard frame and on the light table to ensure the watermark is in the same position in the camera’s field of view. Another reflected image is captured with the verso facing up.
A multiplicative model is used to model the transmission image acquisition process. The proposed model is used to estimate the watermark image by registering the acquired images, followed by a simple image processing step. The result is a fully automatic algorithm for watermark visualization which does not require image processing expertise to operate.
RESULTS ON ONE INKED SIDE DOCUMENT
 |
 |
 |
|---|---|---|
| (a) | (b) | (c) |
| Fig.2 – (a) Original transmission image. (b) Watermark obtained with the proposed method. (c) Watermark obtained with beta-radiography. | ||
RESULTS ON TWO INKED SIDES DOCUMENT
 |
 |
 |
||||
|---|---|---|---|---|---|---|
| (a) | (b) | (c) | ||||
 |
 |
|||||
| (d) | (e) | |||||
| Fig.2 – (a) Original transmission image. (b) Watermark obtained with the proposed method. (c) Watermark obtained with beta-radiography. (d) Detail of watermark obtained with the proposed method. (e) Detail of watermark obtained with beta-radiography. | ||||||
LINKS OF INTEREST
The manuscript of this work is available in this link. The MATLAB© code to run the proposed algorithm as well as the images used in this work are available in our Github repository github.com/IVPLatNU.