Image Processing Reference
In-Depth Information
visual media and the ease in their storage, and transmission are more and more ex-
ploited to convey information. Together with undoubted benefits, the accessibility
of digital images brings a major drawback. With development of low cost, power-
ful image editing tools, the craft of tampering visual content is no more restricted
to experts, so they can easily change image content and also it's meaning without
leaving any traceable effect.
Image forgery detection methods in computer vision are quite able to authenticate
the entire content of a digital image and protect them against tampering. A reliable
images forgery detection system will be useful in many areas such as surveillance
systems, medical imaging, criminal investigation, journalism, visa and immigration
documents, insurance processing and forensic investigation.
The forgery detection techniques that are developed for digital images are mainly
classified into two major classes, active and passive [2, 5] While in the active meth-
ods we would like to insert data or signature at the time of digitizing, the passive
methods operate in the absence of any data or signature [2, 5]. In the active methods,
we embed data into the original image to protect it against the forgery, where in the
passive methods we don't have the original image and we should investigate some
features such as statistical anomalies, correlations, compressions and measurements
of objects in the existence image to detect forgery [4, 5].
Active approaches can be divided into two categories by the embedding in the
position of spatial domain or frequency domain data [4]. Spatial domain techniques
have already developed and are easier to implement but are limited in robustness [3].
Data embedding in the spatial domain consists of insertion and detection stages. The
insertion algorithms are used to embed the data into the digital image and detection
algorithms extract those data.
On validation and authentication aspects, the data which is embedded in a spatial
domain should be unpredictable, invisible and also sensitive to any modification
[3, 5].
In 2013 Anoop et al. [1] presented a full image encryption algorithm base on
transform domain and stream cipher. In 2009 Krikor et al. [9] used DCT and stream
cipher for digital image encryption. In 2003 Pommer et al. [13] provided an image
encryption approach using selective encryption of wavelet packet. In 2003 Droogen-
broeck et al. [19] developed Triple DES and IDEA based approach for the purpose
of digital image encryption.
Using cellular automata as a discrete model is another way to generate such in-
tricate information. This information would be embedded in a particular domain
of an image for the purpose of image encryption and digital image forgery detec-
tion. In 2013 Xiaoyang et al. [20] used elementary cellular automata state rings to
encrypt and decrypt QR code binary image. In 2012 Jin [8] developed an image en-
cryption approach using the behavior of a number of Elementary Cellular Automata
(ECA) with periodic boundary conditions. In 2011 Malakooti et al. [16] proposed
a method to use the one dimensional cellular automata including statistical infor-
mation of a digital image as an operational and practical way for image forgery
detection. See also chapter 5.6 in this topic which describes a passive method for
copy-move forgery detection using cellular automata.
 
Search WWH ::




Custom Search