This dataset contains documents printed on 7 inkjet and 13 laser printers. Each printer printed approx. 60 unique documents of three different document types, invoices, contracts and scientific literature. The invoice and contract documents were created artificially. The scientific literature consists of real world data crawled from the web. All documents of the scientific literature type originally have been released under a license that allows reusing.
The dataset contains the original PDFs in order to allow reprinting or expanding the dataset, as well as the printed and scanned images.
Additionally the dataset contains the extracted features as CSV files as described in the Thesis Document Authentication using Printing Technique Features
The dataset can be downloaded here: DFKI Printing Technique Dataset (1.96 GB)
Contact: Markus Goldstein