Learning-based and rule-based source color laser printer identification for mobile environment

Abstract

A source printer identification technique can be employed as a countermeasure to forgeries using color laser printers. In this dissertation, two new methods are presented to identify color laser printers

a method using halftone texture fingerprint and a method using deep learning framework. Proposed methods use images photographed without additional close-up lens as input images. In the rule-based proposed method, halftone texture fingerprints are extracted in the curvelet transform domain. The extracted halftone texture fingerprint is used in correlation-based detection, and the color laser printer of the most similar known halftone texture fingerprint is determined as the source color laser printer. The learning-based proposed method is mainly divided into two components

improved halftone color channel decomposition based on a GAN-alike framework, and printer identification based on a CNN. The halftone color decomposing ConvNet is trained with the refined dataset by using S+U learning, and the trained knowledge is transferred to the printer identifying ConvNet to enhance the accuracy. The robustness about rotation and scaling is considered in training process, which is not considered in existing methods.Experiments are performed on eight color laser printers and the performance is compared with several existing methods. The experimental results show that the proposed method outperforms existing source color laser printer identification methods.