Jour of Adv Research in Dynamical & Control Systems, Vol.

10, 12-Special Issue, 2018

Survey of Image Forgery Detection Methods-

a Passive Approach
N. Gladiss Merlin, Assistant Professor, Department of Computer Science and Engineering, Jeppiaar Institute of Technology,
Chennai, Tamilnadu. E-mail:[email protected]
K. Karthika, Assistant Professor, Department of Information Technology, St.Joseph College of Engineering, Chennai,
Tamilnadu. E-mail:[email protected]
Rajan Cristin, Assistant Professor, Department of Computer Science and Engineering, GMR Institute of Technology, Rajam,
Andhra Pradesh. E-mail:[email protected]
K. Karthick, Associate Professor, Department of Electrical and Electronics Engineering, GMR Institute of Technology, Rajam,
Andhra Pradesh. E-mail:[email protected]
Abstract--- Image forgery detection technique aims to authenticate the originality of the image and localize the
tampered region in the image. This area of research is relatively new and only few contributions have been made
that relate to the detection of the image forgeries. Forgery detection is the view point of this research. For any
investigation to be held it is important to know about the existing concepts and methodologies in the corresponding
filed. Only through analysis, the fact for incorporating new concepts can be made. But in the passive approaches, the
category realization is ambiguous. The uncertainty is because of the different types of the forgery practiced over the
images. For every tampering action, a suitable detection approaches is obligatory. In current malicious trend, new
types of forgeries come into existence every day, for the mere action, new detection techniques are being developed
which makes the categorization knotty in such cases, the literature is the viable solution.
Keywords--- Image Forgery, Copy-Move, Shadow, Reflection.

I. Introduction
In recent years, the need for the forgery detection algorithm is increasing because of the rapid growth and
availability of the imaging processing softwares and the advancements made in digital cameras Snigdha K et al[17].
Moreover, the necessity ascends because of the bearing of image originality in different environment such as
forensic investigation, criminal investigation, law enforcement, journalist photography etc. A minute mislead in the
authenticity have greater effect. Moreover in the present information era, the massive images are available in the
internet. For any malicious purpose, the image from the web can be collected and altered for the planned purpose. In
every aspect of society, the forged image is found to be blow-out for misleading or illegal purposes. It is precisely
necessary to analyse these images to establish their authenticity. Either the tampering is done with mislead intention
or not, authenticity is obligatory. Bearing in mind the wide range of fields where the influence of tampered image
seems intolerable, the need for the reliable, efficient technique for image forgery detection is increasing Pearson H
[15], A.C. Popescu and H. Farid[16], Cristin et al [7]. Many of the state of art image forgery detection techniques
have been proposed in recent years.
Literature presents variety of the forgery detection techniques for the originality validation of the image. A
detailed survey has been carried out to identify the various research articles available in the literature in all the
categories of the forgery detection techniques. The major contributions made in the existing works are analysed with
specific mention to the advantages and disadvantages in the literature. In the following sections, the relevant
literature applicable for the valuation of the state-of-art work on the image forgery detection technique is provided.
Based on the survey conducted, in the passive approaches, seven different categories of detection techniques are
discussed. They are Copy-move detection techniques ,Splicing detection techniques, Camera based detection
techniques, Image enhancement based detection techniques, Image format based detection techniques, Shadow
based detection techniques and Reflection based detection techniques
Based on the detailed survey conducted over the image forgery detection techniques, taxonomy is developed for
realising the different approaches. The taxonomy provided in this research is not universally acceptable one.
However, it aids in differentiating the various active and passive approaches of the image forgery detection
technique. Figure 1.1 depicts the literature taxonomy of the image forgery detection techniques.

ISSN 1943-023X 88
Received: 23 July 2018/Accepted: 21 Aug 2018
Jour of Adv Research in Dynamical & Control Systems, Vol. 10, 12-Special Issue, 2018

Image Forgery Detection

Passive
Approaches

Copy-move detection techniques

Splicing based detection Shadow based detection Reflection based detection

techniques techniques techniques

Image enhancing based detection Image format based detection

Camera based detection techniques
techniques techniques

Figure 1.1: Literature Taxonomy

II. Passive Approach based Forgery Detection

Passive approaches are the most expedient forgery detection technique. The passive approaches based forgery
detection technique is also called the blind forgery detection. The blind name factor ascends because of the
following fact that, passive approach only uses the received image for the originality check without any modification
in image at the time of creation or capture. The passive approaches are adopted with the wellbeing assumption that
the tampering action performed in the image affects the statistical property (inconsistency) of the image scene
content which introduces new artifacts resulting in various forms of inconsistencies. By assessing these artifacts, the
forgery is analysed in the passive approaches based detection techniques. The negligence of the prior information of
the image makes it most popular in the current trends, where most of the pictures which are filed in for the
authenticity are collected from the internet. Different types of the passive approaches are available based on the kind
of forgery and the inconsistency twisted. An overview of the existing blind forgery techniques is reviewed in this
section.
Copy-Move Detection Techniques
Copy move forgery is the most shared and straight forward type of the image tampering technique. The
tampering procedure in the copy move forgery is simple because the original parts of the image are copied and
moved to a desired location and pasted. The area duplication facilitates hiding certain details of the image. The part
selected for the duplication is ordinarily the textured regions since it has the similar color and noise variation
properties. Such belongings make the modifications unperceivable for the human eye. The literature of the existing
research works related to the copy move forgery detection technique is given below.
Irene Amerini et al [10] proposed an approaches for detecting the copy move image forgery using SIFT based
method. In the main, to detect the copy move forgery, it is essential to recover the geometric information used to
perform the duplication. This is because the forger can achieve the tampering target only by the application of the
satisfactory geometric transformation. In this method, to detect the modifications made in the image a novel method
based on scale invariant features transform was developed. The SIFT was used in this approaches to identify and
describe the cluster points of the duplicated region. Upon the detection, the cluster points were used to reconstruct
the geometric transformation parameters

ISSN 1943-023X 89
Received: 23 July 2018/Accepted: 21 Aug 2018
Jour of Adv Research in Dynamical & Control Systems, Vol. 10, 12-Special Issue, 2018

Splicing based Detection Techniques

Image splicing is another important and frequent type of forgery handled by the forgers for the image
manipulation. In image splicing, the image fragments from one or more images are jointed into an original image
creating false facts and evidence. The splicing forgery is commonly used because the effort required for modifying
the image is minimal. Because of its simplicity, the image splicing detection is considered as the fundamental task of
the image forgery detection. The splicing of the different image into the original image alters the correlation level
which introduces new artifacts which can be used for detection. The literature of the existing research works related
to the image splicing forgery detection is presented below.
Wei Zhang et al.[18] obtained a splicing detection method based on the planar homography and graph cut
method. The proposed method requires an automatic fake region detection method for identifying the splicing
artifacts and fake region extraction method for localizing the forgery information. Illustrated detection technique
used a planar homography constraint for detecting the fake region in the image which is specified as target. The
target region contains the geometrical constraints which are responsible for the modification. Upon the target
acquire, an extraction method was utilized to extract the fake objects from the manipulated image.
Image Enhancement based Detection Techniques
Image enhancement based forgery detection technique is a distinctive detection methodology concerned with the
detection of tampering action because of the image enhancement process. In general, the forgers tamper the image
more imperceptible by including some image enhancement process such as resizing, resampling, rescaling, rotation,
blurring etc. The enhancement operations performed on the image for the manipulation are often invisible tom
human eye. But the image enhancement process introduces the specific correlation in the image which can be used
as an evidence for the forgery detection. Figure 2.1 depicts three types of image enhancing process. Only three
enhancement processes is signified in the literature since the research works related to the resampling, rescaling and
blur based forgery detection technique appears to be more conversant.

Image enhancing process based techniques

Resampling Rescaling Blur

Figure 2.1: Types of Image Enhancing Process

Image Format based Detection Techniques
Image format based detection technique utilizes the format variation of the image as the evidence for the
tampering detection. On creation of forged image, the resultant image is inherent to different kind of compression
artifacts based on the tampering source. The compression artifacts may be due to the recompression procedure.
These compression artifacts results in variation in compression characteristics of the image which can be used to
validate the image integrity. JPEG is the most export image file format used in most of the digital camera. The
image processing operations applied for the tampering action affects the compression format ascending issues which
is very essential in image tampering detection. The literature of the existing research works related to image format
based forgery detection techniques are given below.
Tiziano Bianchi et al.[2] developed a technique for detecting the image forgery based on image format. The
detection was centred on examining the JPEG double compression. In order to test the presence or absence of the
double compression atrifacts, they have used a unified statistical model which automatically computes the likelihood
map of the DCT blocks of the image. But image pair with double compressed scenario was hard to detect. Another
method centred on double compression artifacts as evidence for tampering detection was proposed by Piva et al[1].
Illustrated algorithm, only detect the nonaligned double JPEG (NA-JPEG) recompression. This method examines
the DCT coefficient feature of the image. Simple threshold detector was used to classify the NA-JPEG images. The
experimentation of their method resulted in superior results in terms of detection accuracy. Focused on the image
artifacts due to the recompression, Yi-Lei Chen et al.[12] proposed a new method for tampering detection. They

ISSN 1943-023X 90
Received: 23 July 2018/Accepted: 21 Aug 2018
Jour of Adv Research in Dynamical & Control Systems, Vol. 10, 12-Special Issue, 2018

have designed a robust detection approaches for discovering all the traces cause by the recompression of the images.
The Periodic characteristic of the JPEG image in spatial and transform domain was utilized to detect the block
aligned and misaligned recompression.
Camera based Detection Techniques
Camera based detection technique is one of the unique forgery detection technique. This detection technique is
concerned with the imaging process using the camera. The digital camera consists of lens system, sampling filter,
color filter array, colour interpolation and post processor. The processing in each of the individual elements of the
camera varies from one another. In perspective of lens system, improper light focus during the capture embraces the
correlation which can be used as evidence for tampering detection. Similarly, in imaging sensor, and CFA, the
possibility for the inbound correlation is increasing which can be used as forgery evidence. Figure 2.2 depicts the
different detection techniques in the camera imaging process well-thought-out in the literature.

Camera Imaging Process based

techniques

Imaging Sensor based

techniques CFA based techniques

Lens System based techniques Sensor noise based techniques

Figure 2.2: Types of Detection Techniques in Camera Imaging Process

Shadow based Detection Techniques
Shadow based detection technique is relevant to the physics based passive approaches (as mentioned in section
1.5.2 of chapter 1. In the image manipulation process, geometric transformations are given to the image to give the
impression that the image is authentic. The transformation include enhancement process, modification process etc.
However, shadow in an image is inevitable. The geometric transformation concerned with the image manipulation in
regard to the shadow is left overlook by the forgers. This shadow inconsistency can be used as an evidence for the
tampering detection. Shadow based detection technique is attractive compared to other detection techniques because
the modification of the shadow present in the image is difficult to attain. Moreover, shadows survive common
compression operation performed for manipulation. The literature of the existing research works related to the
shadow based detection technique is given below,
YongzhenKe et al.[13] investigated a method to detect the image composites by imposing the geometric
constraints from the shadow inconsistency of the image. In particular, the authors explored the suspicion region of
the image including shadow and non-shadow. Illustrated method, mainly concentrated on the image with
inconsistent shadow because of the splicing operation. They reached the goal of image composite exposure by
comparing the features of the image regions with and without shadow. Primarily, suspicious region from the image
was manually selected and masked to separate the shadow and non-shadow region. Consequently, the texture feature
of the shadow and non-shadow region (inside and outside) were extracted. In this method, Local Binary Pattern was
adopted as feature extractor because of its practical significance. The extracted features were then compared using
correlation similarity function and the tampered images are estimated. Moreover, they have introduced a different
technique acceptable for detecting any type of forgery in which the strength of the light source was estimated from
the image and compared by the correlation function for the detection. The experimentation of their method resulted
in forgery detection with enhanced detection accuracy.
Another shadow based detection method utilizing the shading and shadow inconsistency was introduced by the
Eric Kee et al. [14]. They used the physical inconsistency in lighting from the shading as well as shading as the

ISSN 1943-023X 91
Received: 23 July 2018/Accepted: 21 Aug 2018
Jour of Adv Research in Dynamical & Control Systems, Vol. 10, 12-Special Issue, 2018

forgery evidence. This method differs from preceding method for the reason of combined consideration of shading
and shadow. The significant advantage of this method is related to assumptions made on the scene geometry. In the
illustrated method, the multitude of shading and shadows based constraint were estimated on the projected location
of a distance point light source. Shading constraint from the 2-D and 3-D direction of distinct point light was
estimated separately. The estimated constraints were formerly combined with a linear framework to determine the
consistency of the shadow and shading with single light source. The consistency deliberates the originality of the
image. Moreover, with the combined multiple shadow constraint, they have checked the possibility in consistency
detection. This method obtained promising results in visually compelling forgery images.
Reflection based Detection Techniques
Reflection based detection technique is related to the geometry based passive approaches.. When altering an
image, in order to conceal the traces of the forgery various processing operations are performed, but it is often
difficult to exactly match the reflections of the objects present in the image content. The diffusive reflection of the
object can be compelled at the first glance however at close examination the reflection inconsistency can be
discovered. The inconsistency in the reflection can be used as an evidence for the tampering detection. The
inconsistency in shadow can be analysed by estimating the direction of the light source. Like shadow based
detection technique, reflection based detection technique is also a favourable forgery detection technique. The
literature of the existing research works related to the reflection based forgery detection technique is given below.
James F. O’Brien et al [11] described a method for exposing the image manipulation using the inconsistent
reflection in the image. The geometric inconsistencies ascend in the manipulated image at time when the forger tries
to insert the fake reflection of the object in the image scene content. By analysing this geometric inconsistency, the
tampering action performed over the original image can be detected and localized. They have used the basic rules of
the reflective geometry, and linear perspective projection for the reflection analysis. In the illustrated detection
technique, for attaining the reflection information some minimal assumptions are made about the image scene
geometry. Primarily, they tried to obtain the underlying geometric relation in the image by combining the 3-D
geometric relation of planar reflection with the geometry of linear perspective projection. The reflection vanishing
points alongside the orthogonal direction of the objects, reflection line midpoints and centre of projection point,
implausible midpoints corresponding to the feature points of the image were estimated. After the detection of the
reflection point concerned with different constraints, inconsistencies were detected by comparing the reflection point
and the object feature point.
Another image forgery detection method using the inconsistent reflection vanishing point was described by
Huayongge et al. [9] and Cristin and Cyril raj [5][6]. They aimed to reduce the appropriate feature selection
problem which raises the uncertainty in large error distance of reflection vanishing point. In this method, Sin2()
based function was used to normalize the error distance. Moreover, the maximal normalized error distance was
utilized to expose the image forgery. In this method, initially, the reflection vanishing points of the objects in the
seen was estimated directly or indirectly of the planar reflection. The error distance between the reflection point
intersections is measured. The dispersion of the intersection scattering is much larger for the forged image and
smaller for the authentic image. To exploit the forgery, the maximum error distance from the intersection scattering
was found and analysed.

III. Image Processing Techniques for Image Shadow and Reflection Detection
To detect the manipulated image composite with geometric inconsistencies of shadow or reflection, the
discovery of the shadow/reflection region is requisite. In general, many of the image processing techniques is
applied to the image for the separating the image and mesh data Kang, G., [12]. Image processing is a technique in
which the images are processed using mathematical operations using any form of the signal processing. The
processing output is either image or a set of characteristics or parameters regarding the image which embraces the
mesh data. Segmentation is an important technique used in image processing to identify the objects in the image.
Based on the object discovery, the image analysis can be done Cristin and Cyril raj [4][6].
Image segmentation technique is accepted as an applicative task in most of the image science field. In the image
forgery detection, the segmentation is applied to extract the inconsistent region from the image based on which the
features are extracted, matched and analysed to validate the image originality. The block diagram of segmentation
based shadow and reflection detection is depicted in figure 3.1. Figure 3.1 epitomize the common flow of the
segmentation adopted shadow or reflection based forgery detection. Primarily, the image subjected to the originality
validation is subjected to any pre-processing procedures (if necessary). Subsequently, image processing operation

ISSN 1943-023X 92
Received: 23 July 2018/Accepted: 21 Aug 2018
Jour of Adv Research in Dynamical & Control Systems, Vol. 10, 12-Special Issue, 2018

segmentation is applied on to the accepted input image. Image segmentation partitions the digital image into
multiple segments. In the view point of shadow or reflection based forgery detection technique, the analysis will be
made ease only after separating the shadow or reflected region from the image which is provided by segmentation.

Image Image Segmentation Feature Extraction

Region growing Clustering Level Set Threshold based

Morphological
Map based Gray scale Super pixel
analysis

Detection result Similarity analysis Matching

Figure 3.1: Segmentation Technique based Shadow and Reflection Detection

IV. Machine Learning Technique for Image Shadow and Reflection Detection
Artificial intelligence is a ‘state of art’ problem solving entity used in any field associated with the computer
vision. Machine learning technique is a type of artificial intelligence with the ability to learn without being explicitly
programmed and predict or classify based on the learned data Carbonell, J.G., et al.[3]. Thorough data analysis is
feasibly provided by the machine learning techniques. In image forgery detection, machine learning technique is
used for validating the image authenticity. This is because of the composite nature of the objects in the scene
content. In the view point shadow/reflection based forgery detection, the practice of machine learning technique is
indispensible. The learning based detection approaches improves the detection performance of the adapted detection
scheme. Some of the significant advantage of the machine learning based techniques are) knack to solve many real
world problem ii) ability to classify the database with incomplete information. These advantages are obligatory in
the image forgery detection.

Image Image segmentation Feature Extraction

SVM Neural Network

K-NN Naive Bayes Classifier

Fuzzy Decision Tree

Original
Forged Image
Image

Figure 4.1: Machine Learning Technique based Shadow and Reflection Detection Technique
Machine learning technique is accepted as one of the main stays in the image forgery detection. The block
diagram of the machine learning technique for shadow and reflection based forgery detection technique is shown in
figure 4.1. Primarily, the image accepted for the originality check is subjected to the image segmentation. In image
segmentation step, the shadow/reflection region of the image is segmented using any of the segmentation technique.

ISSN 1943-023X 93
Received: 23 July 2018/Accepted: 21 Aug 2018
Jour of Adv Research in Dynamical & Control Systems, Vol. 10, 12-Special Issue, 2018

After region separation, the feature points are extracted from the corresponding shadow or reflection region. The
features more concerned with the image texture, light strength etc. are extracted. Such features are extracted because
only those have the bursting association with shadow/reflection inconsistency. After the feature extraction, the
feature points are fed to the machine learning technique.

V. Conclusion
The requisite for the image forgery detection is increasing to reimburse the forgeries introduced in the modern
world. A survey has been made to identify the research articles concerned with the forgery detection. Based on the
type of forgery performed, literature has been provided in this chapter. The passive approaches based detection
techniques such as copy move detection, splicing detection, Image enhancement detection, image format detection,
camera imaging detection, shadow detection and reflection detection techniques are discussed with explicit remark
on the advantages and shortcomings. As per the literature, the blind approaches with ability to detect the originality
of the image modified with any type of forgery seem necessary. It also describes the segmentation methods for
separating the shadow or reflection region from the image for the forgery detection and the machine learning
techniques for the shadow or reflection based forgery detection is summarized.

References
[1] Bianchi, T. and Piva, A. Detection of Nonaligned double JPEG compression based on integer periodicity
maps. IEEE transactions on Information Forensics and Security 7 (2) (2012) 842-848.
[2] Bianchi, T. and Piva, A. Image forgery localization via block-grained analysis of JPEG artifacts. IEEE
Transactions on Information Forensics and Security 7 (3) (2012) 1003-1017.
[3] Carbonell, J.G., Michalski, R.S. and Mitchell, T.M. An overview of machine learning. Machine learning 4
(4) (1983) 3-23.
[4] Cristin, R. and Raj, V.C. Exposing Image Manipulation with Consistent Shadows Using Vanishing points.
International Journal of Advancements in Computing Technology 7 (3) (2015) 79-89.
[5] Cristin, R. and Raj, V.C. Exposing Image Manipulation with Curved Surface Reflection. Indian Journal of
Sciences & Technology 9 (38) (2016) 01-09.
[6] Cristin, R. and Raj, V.C. Consistency features and fuzzy-based segmentation for shadow and reflection
detection in digital image forgery. Science China Information Sciences 60 (8) (2017).
[7] Cristin, R., Ananth, J.P. and Raj, V.C. Illumination-based texture descriptor and fruitfly support vector
neural network for image forgery detection in face images. IET Image Processing (2018).
[8] Fan, N., Jin, C. and Huang, Y. A pixel-based digital photo authentication framework via demosaicking
inter-pixel correlation. Proceedings of the 11th ACM workshop on Multimedia and security, 2009, 125-130.
[9] Ge, H. and Malik, H. Exposing image forgery using inconsistent reflection vanishing point. International
Conference on Audio, Language and Image Processing (ICALIP), 2014, 282-286.
[10] Amerini, I., Ballan, L., Caldelli, R., Del Bimbo, A. and Serra, G. A sift-based forensic method for copy–
move attack detection and transformation recovery. IEEE Transactions on Information Forensics and
Security 6 (3) (2011) 1099-1110.
[11] James, O. and Farid, H. Exposing photo manipulation with inconsistent reflections. proceedings of ACM
Transaction on Graphics 31 (13) (2012) 1-11.
[12] Kang, G., 1977. Digital image processing. Quest, Autumn (1977).
[13] Ke, Y., Qin, F., Min, W. and Zhang, G. Exposing image forgery by detecting consistency of shadow. The
Scientific World Journal 12 (2) (2014).
[14] Kee, E., O'brien, J.F. and Farid, H. Exposing Photo Manipulation from Shading and Shadows. ACM Trans.
Graph 33 (5) (2014).
[15] Pearson, H. Image manipulation: CSI. cell biology Nature 434 (6) (2005) 952-953.
[16] Popescu, A.C. and Farid, H. Exposing digital forgeries by detecting duplicated image regions. Technical
Report, TR-515 23 (12) (2004) 1224-1236.
[17] Mankar, S.K. and Gurjar, A.A. Image Forgery Types and Their Detection: A Review. International Journal
of Advanced Research in Computer Science and Software Engineering 5 (4) (2015) 174-178.
[18] Wei, W., Wang, S., Zhang, X. and Tang, Z. Estimation of image rotation angle using interpolation-related
spectral signatures with application to blind detection of image forgery. IEEE Transactions on Information
Forensics and Security 5 (3) (2010) 507-517.

ISSN 1943-023X 94
Received: 23 July 2018/Accepted: 21 Aug 2018