Intelligent Printing Technique Recognition and Photocopy Detection for Forensic Document Examination
The detection of fraudulent documents is becoming an important issue in large scale office automation. In this context the recognition of a documents creation process is of valuable information in the examination of a questioned documents authenticity. We propose a novel method of printing technique recognition and photocopy detection based on statistical supervised machine learning. This is achieved by the classification of statistical document features that are obtained by an analysis of a given documents spatial and frequency domain. For the purpose of testing our methods we prepared a representative database of documents. Our results showed that (1) most of the documents are classified correctly (2) even at low scan resolutions.