Source Camera Identification Forensics Based on Wavelet Features

Abstract

Source camera identification forensics aims at determining and authenticating the original sources of digital images to support forensics and get the trace of digital images. This paper introduces a new wavelet features based passive forensic method for the identification of the image source camera. We consider the intrinsic defects and processing of imaging pipeline within digital cameras can be used to formulate the source camera identification problem. Based on this idea, we extract higher-order wavelet features and wavelet coefficient co-occurrence features from taken images, and then apply sequential forward feature selection (SFFS) method to reduce the redundancy and correlation of features and finally use multi-class support vector machine (multi-class SVM) as classifier to identify source cameras. The effectiveness of the proposed approach, also in comparison with other approaches, is experimentally proved on images of six digital cameras.