Image forgery detection and its accuracy are addressed in the proposed work. The image authentication process aims at finding the originality of an image. Due to the advent of many image editing software image tampering has become common. The Enhanced hashing approach is suggested for image authentication. The concept of Hashing has been used for searching images from large databases. It can also be applied to image authentication as it produces different results with respect to the change in image. But the hashing methods used for similarity searches cannot be used for image authentication since they are no sensitive for small changes. Moreover, we need a system that detects only perceptual changes. A new hashing method, namely, enhanced robust hashing is proposed for image authentication, which uses global and local properties of an image. This method is developed for detecting image forgery, including removal, insertion, and replacement of objects, and abnormal color modification, and for locating the forged area. The local models include position and texture information of object regions in the image. The hash mechanism uses secret keys for encryption and decryption. IP tracing is done to track the suspicious nodes.
Enhanced Hashing Approach For Image Forgery Detection With Feature Level Fusion
1. INTERNATIONAL JOURNAL FOR TRENDS IN ENGINEERING & TECHNOLOGY
VOLUME 5 ISSUE 2 – MAY 2015 - ISSN: 2349 - 9303
192
Enhanced Hashing Approach For
Image Forgery Detection With
Feature Level Fusion
G. Mathumitha
PG Scholar, Department of CSE,
Paavai College of Engineering,
Namakkal, India.
R. Murugesan
Assistant Professor, Department of CSE,
Paavai College of Engineering,
Namakkal, India.
Abstract—Image forgery detection and its accuracy are addressed in the proposed work. The image authentication process aims
at finding the originality of an image. Due to the advent of many image editing software image tampering has become common.
The Enhanced hashing approach is suggested for image authentication. The concept of Hashing has been used for searching
images from large databases. It can also be applied to image authentication as it produces different results with respect to the
change in image. But the hashing methods used for similarity searches cannot be used for image authentication since they are no
sensitive for small changes. Moreover, we need a system that detects only perceptual changes. A new hashing method, namely,
enhanced robust hashing is proposed for image authentication, which uses global and local properties of an image. This method is
developed for detecting image forgery, including removal, insertion, and replacement of objects, and abnormal color
modification, and for locating the forged area. The local models include position and texture information of object regions in the
image. The hash mechanism uses secret keys for encryption and decryption. IP tracing is done to track the suspicious nodes.
Index Terms—Image forgery, image hashing, global and local properties, perceptual hashing, image authentication
—————————— ——————————
1 INTRODUCTION
Digital images are increasingly transmitted over non-secure
channels such as the Internet. Therefore, military, medical and
quality control images must be protected against security attacks.
Hence, image authentication has become a mandatory process in
image sharing. An image hash function maps an image to a short
binary string based on the image's appearance to the human eye.
With advancement in technology, there are many multimedia data
available over the internet. As storage becomes less costly, all the
data are stored in database as blob objects.
One primitive way for dealing with massive multimedia
databases is the similarity search problem. It aims to retrieve
similar objects to the query object from the database. Particularly,
similarity search is at the heart of many multimedia applications,
such as image retrieval, video recommendation, event detection,
and face recognition. To improve the performance of similarity
search, a long stream of research efforts has been made in the
database community.
Because of the difference in dimensionality it is difficult to
find the exact image using similarity search. To address this issue
approximate similarity search has been implemented in recent
years, which brings related images as a result instead of exact
images for the given query. With the advent of many image
editing software and its widespread use, image authentication
becomes important to avoid image forgery. Hashing can be
efficiently used to authenticate an image since a small change in
the image will produce a different hash code when the same hash
function is used.
In general, a hash should be short, robust against simple
image modifications and sensitive against major modifications.
Therefore the objective is to provide a reasonably short hash code
for an image with good performance. Global moments of the
luminance and chrominance components are used to reflect the
image’s global characteristics, and extract local texture features
from salient regions in the image to represent the contents in the
corresponding areas.
2 PROPOSED IMAGE AUTHENTICATION PROTOCOL
Many previous schemes are either based on global or local
features. Global features are generally short but insensitive to
changes of small areas in the image, while local features can
reflect regional modifications but usually produce longer hashes.
Therefore, a method that generates reasonably short hash code
and better reflects the properties of an image is required. The
proposed work focuses on efficient and automatic techniques to
2. INTERNATIONAL JOURNAL FOR TRENDS IN ENGINEERING & TECHNOLOGY
VOLUME 5 ISSUE 2 – MAY 2015 - ISSN: 2349 - 9303
193
identify and verify the contents of digital images. The services
provided by the proposed image authentication system are
mentioned below.
1. Identify the received image as a similar image, or a
tampered image, or a different image.
2. Evaluate similarity of two images by calculating the
distance between them.
3. Identify and locate three types of tampered area: Added
area, Removed area, Changed area.
4. Estimate the percentage of tampered area.
Fig. 1. Process steps for image authentication.
When an image is sent to the user, a possible solution to prove
authenticity is to generate a hash value and send it securely to the
user. The hash value is a compact string. It can be called as an
abstract of the content. A user can regenerate hash value from the
received image, and compare it with the original hash value. If
they match, the content is considered as authentic. In order to
allow incidental distortion, the hash value must possess some
robustness.
In [1], the authors generated image hash using Zernike
Moments and local features. But the Salient region is detected as
rectangular boxes which include some background details and
does not clearly show the salient region. In order identify the
salient region edge detection mechanism is used in the proposed
work. And also IP tracing is enabled in the proposed system to
find the malicious node where the image got tampered. Fig.3
explains the process steps of the proposed image authentication
protocol.
The image is first rescaled to a fixed size and converted from
RGB to grayscale. These steps are covered under the
preprocessing step. The aim of rescaling is to ensure that the
generated image hash has a fixed length and the same
computation complexity. Next global and local features are
extracted. Then the Global and local features are concatenated to
construct a final hash value.
2.1 Edge Detection Mechanisms
Our proposed work includes research on the embedding
algorithm robust to geometric distortions and improving the
precision in locating the altered areas by implementing via any
digital multimedia networking application for verifying the
content of image transmission over RGB features.
So this kind of implementation is desired to find features that
best represent the image contents so as to enhance the hash’s
sensitivity to small area tampering while maintaining short hash
length, good robustness against normal image processing like
edge detection mechanisms and include tracer routing to detect
the content modified hacker system which is use full to reduce the
hacking possibilities. So without knowledge of this method,
hacker information may be acknowledged to the sender once the
hacker receives the packet for content or object modifications.
Fig. 2. Salient feature identification using Edge detection.
Here, in the final image salient features are highlighted.
Instead of rectangular boxes only the edges were traced thus
giving only the salient feature. Hash can be constructed for the
detected regions and transmitted along with the original image to
the receiver.
2.2 Hashing Generation and Encryption
The global and object local vectors are concatenated to form
an intermediate hash, which is then pseudo-randomly scrambled
based on a secret key to produce the final hash sequence.
Advanced encryption algorithms are used to encrypt the hash
sequence with respect to secret keys.
Received
Image
Preprocessing
Edge detection
Hash
construction
Fake image Original Image
IP tracing
End Process
Malicious Node Identified
Hash
Function
Match with
original
hash
YesNo
3. INTERNATIONAL JOURNAL FOR TRENDS IN ENGINEERING & TECHNOLOGY
VOLUME 5 ISSUE 2 – MAY 2015 - ISSN: 2349 - 9303
194
The user regenerates the hash value from the received image
after successfully decrypting it and compares it with the original
hash value. If they match, the content is considered as authentic.
Otherwise the received image is identified as a fake one.
2.3 Hamming Distance Matching
Distance between hashes of an image pair is used as a metric
for finding similarity or dissimilarity of the two images. The hash
sequence of a received image needs to be tested with the
decrypted hash sequence under similarity ratio. If the difference is
above the threshold then it has been maliciously tampered else it
is considered as legitimate image.
2.4 IP Tracing
Tracer routing is used to find out the unauthorized router or
the system that purposefully modified the content of the image
and forwarded it to the destination in a routing process. This can
be done by getting acknowledgment from every router in the
routing process by a source. Then source compares the
acknowledgement with the predefined routing table for any
routing delay or IP mismatch. Thus the attacker node can be
identified and reported.
3 DISCUSSION AND CONCLUSION
The proposed image hashing approach is developed using
both global and local features. Image hashes produced with the
proposed method are robust against common image processing
operations like brightness adjustment, rescaling and addition of
noise. The IP tracing mechanism helps to find the malicious node
thus providing a full fledged authentication mechanism.
REFERENCES
[1] Robust Hashing for Image Authentication Using Zernike
Moments and Local Features Yan Zhao, Shuozhong Wang,
Xinpeng Zhang, and Heng Yao, Member, IEEE.
[2] S. Xiang, H. J. Kim, and J. Huang, ―Histogram-based image
hashing scheme robust against geometric deformations,‖ in
Proc. ACM Multimedia and Security Workshop, New York,
2007, pp. 121–128.
[3] V. Monga, A. Banerjee, and B. L. Evans, ―A clustering
based approach to perceptual image hashing,‖ IEEE Trans.
Inf. Forensics Security, vol. 1, no. 1, pp. 68–79, Mar. 2006.
[4] Robust Hashing with Local Models for Approximate
Similarity Search Jingkuan Song, Yi Yang, Xuelong Li,
Fellow, IEEE, Zi Huang, and Yang Yang
[5] Bohm C., Berchtold S., and Keim D. A.(2001), ―Searching in
high-dimensional spaces: Index structures for improving the
performance of multimedia databases,‖ ACM Comput.
Survey, vol. 33, no. 3, pp. 322–373.
[6] Cappelli R. (2011), ―Fast and accurate fingerprint indexing
based on ridge orientation and frequency,‖ TSMCB, vol. 41,
no. 6, pp. 1511–1521.
[7] Datar M. and Indyk P. (2004), ―Locality-sensitive hashing
scheme based on p-stable distributions,‖ in Proc. SCG, pp.
253–262.
[8] Datta R., Joshi D., Li J., and Wang J. Z. (2008), ―Image
retrieval: Ideas, influences, and trends of the new age,‖ ACM
Comput. Survey, vol. 40.
[9] Gionis A., Indyk P., and Motwani R. (1999), ―Similarity
search in high dimensions via hashing,‖ in Proc. VLDB, pp.
518–529.
[10] Jagadish H. V., Ooi B. C., Tan K. L., Yu C., and Zhang R.
(2005), ―iDistance: An adaptive B+-tree based indexing
method for nearest neighbor search,‖ ACM TODS, vol. 30,
no. 2, pp. 364–397.
[11] Jingkuan Song, Yi Yang, Xuelong Li, Fellow, IEEE, Zi
Huang, and Yang Yang (2014) "Robust Hashing With Local
Models for Approximate Similarity search" IEEE
transactions on cybernetics, vol. 44, no. 7
[12] Lv Q., Josephson W., Wang Z., Charikar M. and Li
K.(2007), ―Multi-probe LSH: Efficient indexing for high-
dimensional similarity search,‖ in Proc. VLDB, pp. 950–961.
[13] Salakhutdinov R. and Hinton G. E. (2009), ―Semantic
hashing,‖ Int. J. Approx. Reasoning, vol. 50, no. 7, pp. 969–
978.
[14] Shen H. T., Ooi B. C. and Zhou X. (2005), ―Towards
effective indexing for very large video sequence database,‖
in Proc. SIGMOD, pp. 730–741.
[15] Tao Y., Yi K., Sheng C., and Kalnis P. (2010), ―Efficient and
accurate nearest neighbor and closest pair search in high-
dimensional space,‖ ACM TODS, vol. 35, no. 3.
[16] Weiss Y., Torralba A., and Fergus R. (2008), ―Spectral
hashing,‖ in Proc. NIPS, pp. 1753–1760.
[17] Yang Y., Xu D., Nie F., Luo J. and Zhuang Y. (2009),
―Ranking with local regression and global alignment for
cross media retrieval,‖ in Proc. ACM Multimedia, pp. 175–
184.
[18] Zhang D., Wang J., Cai D. and Lu J. (2010), ―Self-taught
hashing for fast similarity search,‖ in SIGIR, pp. 18–25.
[19] Zhang L., Wang L., and Lin W. (2012), ―Generalized biased
discriminant analysis for content-based image retrieval,‖
TSMCB, vol. 42, no. 1, pp. 282–290.