e-Counterfeit: a mobile-server platform for document counterfeit detection Albert Berenguel *† , Oriol Ramos Terrades * , Josep Llad´ os * and Cristina Ca˜ nero † * Computer Vision Center, Universitat Autonoma of Barcelona, Barcelona, Spain Email: {aberenguel,oriolrt,josep}@cvc.uab.es † ICAR Vision Systems S.A, Barcelona, Spain Email: cristina@icarvision.com Abstract—This paper presents a novel application to detect counterfeit identity documents forged by a scan-printing op- eration. Texture analysis approaches are proposed to extract validation features from security background that is usually printed in documents as IDs or banknotes. The main contribution of this work is the end-to-end mobile-server architecture, which provides a service for non-expert users and therefore can be used in several scenarios. The system also provides a crowdsourcing mode so labeled images can be gathered, generating databases for incremental training of the algorithms. I. I NTRODUCTION Making a fake passport is easy, making a good fake passport is very, very hard. Probably there are few criminal organiza- tions in the world which can produce a counterfeit visa or passport good enough to fool professional passport control. Counterfeit detection has traditionally been a task for law enforcement agencies. EUROPOL and INTERPOL central of- fices are combating document and banknote counterfeiting [1], [2]. They have destined millions of euros in funds to provides technical databases, forensic support, training and operational assistance to its member countries. Despite all these efforts counterfeit detection remains an open issue. There are many different strategies used to fake an identity document (ID) like the alteration of a real passport, impersonation of the legitimate owner or printing false information on a stolen blank real paper, to cite some. An important security feature that serves against counterfeiting and manipulation of documents is the background/security printing. The security printing can be classified into two categories: 1) Security printing techniques and printing processes used: microprinting as well as security inks (e.g. optical vari- able ink, ultraviolet or infrared ink). This security features can only be checked with specialized equipment like ultraviolet lamp or a magnified glass, therefore are not possible to check if our target of image acquisition is the general smartphones cameras. 2) Print designs and security elements: guilloches or fine- line patterns, rainbow coloring, etc., discernible to the naked eye. We focus in this category of elements. The counterfeiter requires high technical specialized print- ing equipment to reproduce this background printing tech- niques. Having this equipment is not feasible for the majority of the counterfeiters due economic and restricted availability issues. Thanks to this, a large part of counterfeits just follows the procedure of scan a real document, alter the data and then print the document with a common commercial printer. We refer to this methodology of counterfeit document generation as scan-printing procedure. Following the scan-printing pro- cedure, we expect that the background print design will loose detail hence it will be possible to classify it as counterfeit. We point our efforts to detect this kind of low quality counterfeit documents. It is important to note that most of the print design methodology that is possible to find in the security documents, were first created and used to detect counterfeit money banknotes. The system proposed in this work has been designed for ID documents, however, as mentioned before, it is generic enough so it can be used for other documents with security background texture as banknotes. The main functionality is the analysis of the document authenticity from a single image using a mobile phone camera within a non- controlled environment. This process is used by services or products that require a genuine identification of the client, such as renting a car, opening a bank account, applying for a loan, checking-in in an hotel, etc. Once authenticity of the document is validated, the purpose of the platform where the service is integrated can take place (reading and storing the personal information of the document holder). The main contribution is a mobile-server framework to de- tect counterfeit documents. This end-to-end system provides: a low cost solution fit to be used with common smartphones, flexibility for further updates and robust validation methods. The mobile-server framework also intends to address the lack of tools to generate datasets acquired from smartphone devices. We have published the framework code to help future researchers 1 . The remainder of the paper is organized as follows. In section II, we review the existing frameworks for banknote counterfeit detection. Section III describes in detail each one of the components of the end-to-end framework proposed and IV explains in detail the Counterfeit module. In section V the dataset is presented. At section VI we explain the current set- up and the discussion of the results. Finally, in section VII we draw some conclusions and outline rewarding avenues for future work. 1 https://github.com/gitabcworld/e-Counterfeit. arXiv:1708.06126v1 [cs.CV] 21 Aug 2017