Research Article
A Feature-Based Forensic Procedure for
Splicing Forgeries Detection
Irene Amerini,
1
Rudy Becarelli,
1
Roberto Caldelli,
1,2
and Matteo Casini
1
1
Media Integration and Communication Center (MICC), Universit` a degli Studi di Firenze, 50134 Firenze, Italy
2
National Interuniversity Consortium for Telecommunications (CNIT), 43124 Parma, Italy
Correspondence should be addressed to Roberto Caldelli; roberto.caldelli@unif.it
Received 4 June 2015; Revised 11 November 2015; Accepted 20 December 2015
Academic Editor: Fazal M. Mahomed
Copyright © 2015 Irene Amerini et al. Tis is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Nowadays, determining if an image appeared somewhere on the web or in a magazine or is authentic or not has become crucial.
Image forensics methods based on features have demonstrated so far to be very efective in detecting forgeries in which a portion of
an image is cloned somewhere else onto the same image. Anyway such techniques cannot be adopted to deal with splicing attack,
that is, when the image portion comes from another picture that then, usually, is not available anymore for an operation of feature
match. In this paper, a procedure in which these techniques could also be employed will be shown to get rid of splicing attack by
resorting to the use of some repositories of images available on the Internet like Google Images or TinEye Reverse Image Search.
Experimental results are presented on some real case images retrieved on the Internet to demonstrate the capacity of the proposed
procedure.
1. Introduction
When looking at a digital image, it is quite common to
wonder if it is original or has been counterfeited in some
way. Such a doubt is basically determined by the easiness
with which digital images can be manipulated to change their
content and especially their visual meaning. Te contexts
where doctored pictures could be involved are very disparate;
they could be used in a tabloid or in an advertising poster
or included in a journalistic report but also in a court of law
where digital (sometimes printed) images are presented as
crucial evidences for a trial in order to infuence the fnal
judgement. So, especially in the last case, reliably assessing
image integrity becomes of fundamental importance. Image
forensics specifcally deals with such issues by studying
and developing technological tools which generally permit
determining, by only analyzing a digital photograph (i.e., its
pixels), if that asset has been manipulated or even which
could have been the adopted acquisition device (such an issue
is not relevant to the topic of the present paper). Moreover,
if it has been established that something has been altered, it
could be important to understand in which part of the image
itself such a modifcation occurred, for instance, if a person
or a specifc object has been covered, if an area of the image
has been cloned, if something (i.e., a face or a weapon) has
been copied from another diferent image, or, even more, ifa
mixture of these processes has been carried out. Among the
diferent attacks that can be carried out to modify an image,
two are surely the most important. Te frst one is the splicing
attack which is performed when a portion of an image has
been cut out and, afer having been adapted (e.g., zoomed in
or out, fltered), is inserted into another one to build a new
“fake image.” Te second one is the copy-move attack which is
basically a splicing attack but what is crucial is that the clipped
image portion is pasted somewhere else within the same
image. On the other side, regarding forgeries individuation,
three are the principal classes of detectors studied so far: those
based on double JPEG compression [1–3] adopted to reveal
splicing attack, those based on inconsistent shadows [4],
and fnally those based on local features descriptors (mainly
SIFT—Scale Invariant Feature Transform) [5–8] usually used
to get rid of copy-move attack. A complete overview of
forensic methods for tampering detection is well introduced
in [9]. In particular, features-based methods (based on SIFT
Hindawi Publishing Corporation
Mathematical Problems in Engineering
Volume 2015, Article ID 653164, 6 pages
http://dx.doi.org/10.1155/2015/653164