Abstract
In this paper, we explore transformed spaces, represented by image illuminant maps, to propose a methodology for selecting complementary forms of characterizing visual properties for an effective and automated detection of image forgeries. We combine statistical telltales provided by different image descriptors that explore color, shape, and texture features. We focus on detecting image forgeries containing people and present a method for locating the forgery, specifically, the face of a person in an image. Experiments performed on three different open-access data sets show the potential of the proposed method for pinpointing image forgeries containing people. In the two first data sets (DSO-1 and DSI-1), the proposed method achieved a classification accuracy of 94% and 84%, respectively, a remarkable improvement when compared with the state-of-the-art methods. Finally, when evaluating the third data set comprising questioned images downloaded from the Internet, we also present a detailed analysis of target images.
Highlights
I N A SOCIETY in which social networks became powerful communication tools and are more ubiquitous than ever, it is paramount to design and deploy methods that guarantee the authenticity of the broadcast information
We extensively study different ways to use combinations of different illuminant maps (IM) for different color spaces and examine the most appropriate image descriptors and classifiers to better capture visual properties that might lead to forgery detection
Riess and Angelopoulou’s [6] method comprises four steps: (1) image segmentation grouping regions of approximately the same object color, for generating blocks named superpixels, which are directly illuminated by the investigated light source illuminant and roughly consistent with the physical model of Inverse-Intensity Chromaticity Space; (2) selection of superpixels to be further investigated by the user; (3) estimation of the illuminant color, which is performed twice, one for every superpixel and another one for the selected superpixels; and (4) calculation of the distance from the selected superpixels to the other ones generating a distance map, which is the basis for an expert analysis regarding forgery detection
Summary
I N A SOCIETY in which social networks became powerful communication tools and are more ubiquitous than ever, it is paramount to design and deploy methods that guarantee the authenticity of the broadcast information. In May 2013, the conman Dimitri de Angelis was sentenced to twelve years in prison for deceiving investors using “photoshoped” photos in which he appeared alongside many different prominent people as, for example, former US president Bill Clinton, as Figure 1(a) portrays In another case, dating to November 23rd 2012, a fake photo went viral in the Internet purporting Brazil’s former president Luiz Inácio Lula da Silva beside Rosemary de Noronha, a suspicious gang leader investigated by the Brazilian Federal Police (see Figure 1(b)). CARVALHO et al.: ILLUMINANT-BASED TRANSFORMED SPACES FOR IMAGE FORENSICS input videos as visual rhythms of the Fourier Spectra for highlighting artifacts associated with biometric spoofing in face recognition systems In this vein, in this paper, we propose to use illuminant maps (IM) [6] as a transformed representation space to highlight different types of inconsistencies present in fake images, not detectable in the original image space, and point out possible image forgeries.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
