Local Polar Pattern for Source Color Laser Printer Identification

Abstract

In this paper, a new methodology for the source color laser printer identification using the counterfeit protection system (CPS) pattern is proposed. First, to achieve scale and rotation invariant, a novel Local Polar Pattern (LPP) for CPS pattern representation is designed. Based on the LPP descriptor, a noise robust CPS pattern extraction algorithm is further proposed. Meanwhile, a similarity measurement is defined to measure the similarity between two CPS pattern. At last, a source color laser printer identification system is proposed. To reduce the cost of data storage and computation, the one-shot learning strategy is adopted for CPS pattern identification. We verify the effectiveness of the proposed method on the public CPS pattern dataset. Experimental results demonstrate that our method achieves state-of-the-art performances and superior than deep neural networks.