Identification method of the capture device of an unmanipulated digital image based on Hellinger distance

Abstract

A method based on Hellinger distance and Photo Response Non-Uniformity (PRNU) signal probability density function is proposed to identify the digital image capture device. The method is applied to an unmanipulated digital image (disputed image) and allows to associate it with one of a set of candidates capture devices. This association is performed through the fingerprint imprinted by the capturing device on the disputed image, then this imprinted fingerprint is compared with the fingerprint of each capturing device. The proposed method was implemented in MatlabTM to show its performance and compared against two different methods. The first method, proposed by Goljan et al. (2009), reached a mean similarity percentage of 100%, and the second method, proposed by Quintanar-Reséndiz et al. (2021), reached 99.35%; and the method proposed here reached a mean similarity percentage of 97.68%.