Image Forensic

This paper describes the implementation of a system that addresses the issues in the area of” Forensic Signature Verification”. Two main approaches exist in this field-signature verification and signature identification. Our efforts focus on offline signature ...

With the advent of powerful image editing tools, manipulating images and changing their content is becoming a trivial task. Now, you can add, change or delete significant information from an image, without leaving any visible signs of such tampering. With more than several millions pictures uploaded daily to the net, the move towards paperless workplaces, and the introduction of e-Government services everywhere, it is becoming important to develop robust detection methods to identify image tampering operations and validate the credibility of digital images. This led to major research efforts in image forensics for security applications with focus on image forgery detection and authentication. The study of such detection techniques is the main focus of this paper. In particular, we provide a comprehensive survey of different forgery detection techniques, complementing the limitations of existing reviews in the literature. The survey covers image copy-move forgery, splicing, forgery due to resampling, and the newly introduced class of algorithms, namely image retouching. We particularly discuss in detail the class of pixel-based techniques which are the most commonly used approaches, as these do not require any a priori information about the type of tampering. The paper can be seen as a major attempt to provide an up-to-date overview of the research work carried in this all-important field of multimedia.

Studies towards image forensics are about a decade old and various forms of research techniques have been presented till date towards image forgery detection. Majority of the existing techniques deals with identification of tampered regions using different forms of research methodologies. However, it is still an open-end question about the effectiveness of existing image forgery detection techniques as there is no reported benchmarked outcome till date about it. Therefore, the present manuscript discusses about the most frequently addressed image attacks e.g. image splicing and copy-move attack and elaborates the existing techniques presented by research community to resist it. The paper also contributes to explore the direction of present
research trend with respect to tool adoption, database adoption, and
technique adoption, and frequently used attack scenario. Finally, significant open research gap are explored after reviewing effectiveness of existing techniques.

Abstrak Pesatnya perkembangan teknologi informasi memungkinkan pengguna untuk melakukan pengolahan citra digital dengan menggunakan software editing tools yang banyak tersedia saat ini dengan sangat mudah. Pengolahan citra digital yang mudah ini tak jarang disalahgunakan oleh segelintir orang untuk melakukan kejahatan seperti pemalsuan citra untuk dijadikan hoax dan disebarluaskan. Permasalahan tersebut membuat penulis terdorong untuk menganalisa keaslian sebuah citra digital dengan menggunakan tools image forensic berupa FotoForensic, Ghiro, Forensically, dan Jpegsnoop. Teknik yang digunakan penulis adalah dengan teknik Error Level Analysis (ELA), Metadata, dan JPEG Compression. Kata Kunci : Citra digital, error level analysis, metadata, JPEG Compression Abstarct The rapid development of information technology using digital image processing with software editing tools that are widely available today is very easy to find. This easy digital image processing is not infrequently done by a handful of people to commit crimes such as falsification of images to be deceived and disseminated. These problems make the writer compelled to analyze digital images by using forensic drawing tools such as FotoForensic, Ghiro, Forensically, and Jpegsnoop. The technique used by the author is the Error Level Analysis (ELA), Metadata, and JPEG Compression techniques.

Due to the ease of access to platforms that can be used by forgers to tamper digital documents, providing automatic tools for identifying forged images is now a hot research field in image processing. This paper presents a novel forgery detection algorithm based on variants of Benford's law. In the proposed method, Mean Absolute Deviation (MAD) feature is extracted using traditional Benford's law. Also, generalized Benford's law is used for mantissa distribution feature vector. In addition to Benford's law-based features, other statistical features are used to construct the final feature vector. Finally, support vector machine (SVM) with three different kernel functions is used to classify original and forged images. The method has been tested on two common image datasets (CASIA V1.0 and V2.0). The experimental results show that 0.27% and 0.21% improvements on CASIA V1.0 and CASIA V2.0 datasets were achieved, respectively in detection accuracy by the proposed method in comparison to best state-of-the-art methods. The proposed efficient algorithm has a simple implementation. Moreover, on the basis of Benford's law rich features are extracted from images so that classification process is efficiently performed by a simple SVM classifier in a short time.

The use of machine-learning for multimedia forensics is gaining more and
more consensus, especially due to the amazing possibilities offered by
modern machine learning techniques. By exploiting deep learning tools, new approaches have been proposed whose performance remarkably exceed those achieved by state-of-the-art methods based on standard machine-learning and model-based techniques. However, the inherent vulnerability and fragility of machine learning architectures pose new serious security threats, hindering the use of these tools in security-oriented applications, and, among them, multimedia forensics. The analysis of the security of machine learning-based techniques in the presence of an adversary attempting to impede the forensic analysis, and the development of new solutions capable to improve the security of such techniques is then of primary importance, and, recently, has marked the birth of a new discipline, named Adversarial Machine Learning. By focusing on Image Forensics and image manipulation detection in particular, this thesis contributes to the above mission by developing novel techniques for enhancing the security of binary manipulation detectors based on
machine learning in several adversarial scenarios. The validity of the proposed solutions has been assessed by considering several manipulation tasks, ranging from the detection of double compression and contrast adjustment, to the detection of geometric transformations and filtering operations.

Rapid penetration of internet and advancements in
communication technology is a paved way to easy access of
digital images. Nevertheless, these advancements also create
ways for malicious users to pirate and sell the copyrighted
content.  Digital watermarking techniques have been deployed
in combating the piracy issue. The current digital
watermarking methods are facing the problems in maintaining
invisibleness, robustness, capacity and security. In this article,
authors are analyzing the various popular algorithms used in
digital watermark of copyright protection and propose a novel
approach by combining three such algorithms aiming to
increase the invisibleness and robustness of the watermarked
image. The experimental results, exemplifies much increase in
invisibleness and a nominal increase in robustness.

Last-generation GAN models allow to generate synthetic images which are visually indistinguishable from natural ones, raising the need to develop tools to distinguish fake and natural images thus contributing to preserve the trustworthiness of digital images. While modern GAN models can generate very high-quality images with no visibile spatial artifacts, reconstruction of consistent relationships among colour channels is expectedly more difficult. In this paper, we propose a method for distinguishing GAN-generated from natural images by exploiting inconsistencies among spectral bands, with specific focus on the generation of synthetic face images. Specifically, we use cross-band co-occurrence matrices, in addition to spatial co-occurrence matrices, as input to a CNN model, which is trained to distinguish between real and synthetic faces. The results of our experiments confirm the goodness of our approach which outperforms a similar detection technique based on intra-band spatial co-occurrences only. The performance gain is particularly significant with regard to robustness against post-processing, like geometric transformations, filtering and contrast manipulations.

Copy–move forgery is one of the most common types of tampering for digital images. Detection methods
generally use block-matching approaches, which first divide the image into overlapping blocks and then extract and compare features to find similar ones, or point-based approaches, in which relevant keypoints are extracted and matched to each other to find similar areas. In this paper, we present a very novel hybrid approach, which compares triangles rather than blocks, or single points. Interest points are extracted from the image, and objects are modeled as a set of connected triangles built onto these points. Triangles are matched according to their shapes (inner angles), their content (color information), and the local feature vectors extracted onto the vertices of the triangles. Our methods are designed to be robust to geometric
transformations. Results are compared with a state-of-the-art block matching method and a point-based method. Furthermore, our data set is available for use by academic researchers.

Amidst the continual march of technology, we find ourselves relying on digital videos to proffer visual evidence in several highly sensitive areas such as journalism, politics, civil and criminal litigation, and military and intelligence operations. However, despite being an indispensable source of information with high evidentiary value, digital videos are also extremely vulnerable to conscious manipulations. Therefore, in a situation where dependence on video evidence is unavoidable, it becomes crucial to authenticate the contents of this evidence before accepting them as an accurate depiction of reality.

Digital videos can suffer from several kinds of manipulations, but perhaps, one of the most consequential forgeries is copy-paste forgery, which involves insertion/removal of objects into/from video frames. Copy-paste forgeries alter the information presented by the video scene, which has a direct effect on our basic understanding of what that scene represents, and so, from a forensic standpoint, the challenge of detecting such forgeries is especially significant. In this paper, we propose a sensor pattern noise based copy-paste detection scheme, which is an improved and forensically stronger version of an existing noise-residue based technique. We also study a demosaicing artifact based image forensic scheme to estimate the extent of its viability in the domain of video forensics. Furthermore, we suggest a simplistic clustering technique for the detection of copy-paste forgeries, and determine if it possess the capabilities desired of a viable and efficacious video forensic scheme. Finally, we validate these schemes on a set of realistically tampered MJPEG, MPEG-2, MPEG-4, and H.264/AVC encoded videos in a diverse experimental set-up by varying the strength of post-production re-compressions and transcodings, bitrates, and sizes of the tampered regions. Such an experimental set-up is representative of a neutral testing platform and simulates a real-world forgery scenario where the forensic investigator has no control over any of the variable parameters of the tampering process. When tested in such an experimental set-up, the four forensic schemes achieved varying levels of detection accuracies and exhibited different scopes of applicabilities. For videos compressed using QFs in the range 70–100, the existing noise residue based technique generated average detection accuracy in the range 64.5%–82.0%, while the proposed sensor pattern noise based scheme generated average accuracy in the range 89.9%–98.7%. For the aforementioned range of QFs, average accuracy rates achieved by the suggested clustering technique and the demosaicing artifact based approach were in the range 79.1%–90.1% and 83.2%–93.3%, respectively.

Due to the ease of access to platforms that can be used by forgers to tamper digital documents, providing automatic tools for identifying forged images is now a hot research field in image processing. This paper presents a novel forgery detection algorithm based on variants of Benford's law. In the proposed method, Mean Absolute Deviation (MAD) feature is extracted using traditional Benford's law. Also, generalized Benford's law is used for mantissa distribution feature vector. In addition to Benford's law-based features, other statistical features are used to construct the final feature vector. Finally, support vector machine (SVM) with three different kernel functions is used to classify original and forged images. The method has been tested on two common image datasets (CASIA V1.0 and V2.0). The experimental results show that 0.27% and 0.21% improvements on CASIA V1.0 and CASIA V2.0 datasets were achieved, respectively in detection accuracy by the proposed method in comparison to best state-of-the-art methods. The proposed efficient algorithm has a simple implementation. Moreover, on the basis of Benford's law rich features are extracted from images so that classification process is efficiently performed by a simple SVM classifier in a short time.

Digital image tampering becomes a common information falsification trend. Copy-Move forgery is one of the tampering types that are used. Image forgery is the science of detecting image tempering whether with a previous knowledge about the source image (active) or without (passive). In this paper, we propose a method which is efficient and fast for detecting Copy-Move regions even when the copied region was undergone rotation modify in spatial domain. The proposed method accelerates blocking matching strategy by parallel comparing between blocks. Firstly, the image is divided into fixed-size overlapping blocks then features are extracted for each block. k-means clustering technique is used to cluster the blocks into different cluster. The feature vectors of each cluster blocks are lexicographically sorted by radix sort, and then a similarity measure is calculated between each nearby blocks to determine their similarity. The experimental results show that the proposed method can detect the duplicated regions efficiently even when an image was modified by jpeg compression, rotation and smoothing conditions. The proposed system reduced processing time up to 75% of other previous works.

Determining the authenticity of an image is now an important area of research .In our work,we attempt to classify whether a digital image is a genuine image or is a manipulated version of some authentic image.We have applied three different algorithms discussed in [1] and [2] with modifications of our own. We have then implemented two techniques discussed in [5] and [6] to detect the fake regions in the image. Our approach is invariant to the type of manipulation that has been done to the images.From our methods,we are able to classify nearly 80% manipulated images.

The study of facial growth is explored in many fields of science, including anatomy, genetics, and forensics. In the field of forensics, it acts as a valuable tool for combating child pornography. The present research proposes a new method, based on relative measurements and fixed references of the human face-specifically considering measurements of the diameter of the iris (iris ratio)-for the analysis of facial growth in association with age in children and sub-adults. The experimental sample consisted of digital photographs of 1000 Brazilian subjects, aged between 6 and 22 years, distributed equally by sex and divided into five specific age groups (6, 10, 14, 18, and 22 year olds ± one month). The software package SAFF-2D® (Forensic Facial Analysis System, Brazilian Federal Police, Brazil) was used for positioning 11 landmarks on the images. Ten measurements were calculated and used as fixed references to evaluate the growth of the other measurements for each age group, as well t...

With the tremendous growth and usage of digital images nowadays, the integrity and authenticity of digital content is becoming increasingly important, and a growing concern to many government and commercial sectors. Image Forensics, based on a passive statistical analysis of the image data only, is an alternative approach to the active embedding of data associated with Digital Watermarking. Benford's Law was first introduced to analyse the probability distribution of the 1 st digit (1-9) numbers of natural data, and has since been applied to Accounting Forensics for detecting fraudulent income tax returns . More recently, Benford's Law has been further applied to image processing and image forensics. For example, Fu et al. proposed a Generalised Benford's Law technique for estimating the Quality Factor (QF) of JPEG compressed images. In our previous work, we proposed a framework incorporating the Generalised Benford's Law to accurately detect unknown JPEG compression rates of watermarked images in semi-fragile watermarking schemes. JPEG2000 (a relatively new image compression standard) offers higher compression rates and better image quality as compared to JPEG compression. In this paper, we propose the novel use of Benford's Law for estimating JPEG2000 compression for image forensics applications. By analysing the DWT coefficients and JPEG2000 compression on 1338 test images, the initial results indicate that the 1 st digit probability of DWT coefficients follow the Benford's Law. The unknown JPEG2000 compression rates of the image can also be derived, and proved with the help of a divergence factor, which shows the deviation between the probabilities and Benford's Law. Based on 1338 test images, the mean divergence for DWT coefficients is approximately 0.0016, which is lower than DCT coefficients at 0.0034. However, the mean divergence for JPEG2000 images compression rate at 0.1 is 0.0108, which is much higher than uncompressed DWT coefficients. This result clearly indicates a presence of compression in the image. Moreover, we compare the results of 1 st digit probability and divergence among JPEG2000 compression rates at 0.1, 0.3, 0.5 and 0.9. The initial results show that the expected difference among them could be used for further analysis to estimate the unknown JPEG2000 compression rates.

Data and information security technique is a way to keep an object confidential. eXchangable Image File Format (EXIF) Metadata is data information about digital image. In the digital forensic investigation, this is very useful for copyright security or safeguarding digital evidence. Steganography and cryptography techniques have been widely used to overcome these problem. However, from the research that has been done in terms of security EXIF of metadata digital image photography, there are still some weaknesses. Among them is the depreciation of the digital image size after decryption. Since the standard EXIF metadata treatmens released by JEIDA and CIPA have different characteristic with JFIF metadata. A null value will be deemed to be the content with a value of 1 in EXIF standard. While in JFIF metadata model, the value of null is considered to be worth 0 (zero). This is make the EXIF metadata security process is shrinking of-25,15% from the original size. With the optimizing null value before encryption, the same metadata will be obtained between original image and after decryption. The null value will be replaced with an empty string (it's worth 1) before encryption. Results from the model obtained an average of-1,51% depreciation of the original digital image size.Therefore, this technique can optimize the digital image copyright protection. I. PENDAHULUAN Citra digital merupakan salah satu jenis barang bukti digital yang memiliki tingkat resiko manipulasi dan kehilangan informasi yang sangat tinggi. Data citra digital (metadata) memiliki informasi yang sangat kompleks, meliputi informasi perangkat pengambilan gambar sampai dengan waktu kapan diambil (Zhou et al., 2016). Sehingga dapat dikatakan metadata citra digital tersebut dapat digunakan sebagai hak cipta. Dalam ranah digital forensik, selain konten citra, metadata dapat digunakan sebagai barang bukti digital di persidangan. Mudahnya manipulasi dari sisi metadata dan konten citra inilah yang menjadi celah untuk melemahkan kedudukan citra digital sebagai alat bukti di persidangan. Banyak sekali cara untuk mengamankan metadata, mulai dari teknik steganografi, kriptografi sampai dengan watermarking. Kombinasi pengamanan eXchangable Image File Format (EXIF) metadata dengan teknik steganografi dan kriptografi juga telah dilakukan penelitian dengan hasil hampir sempurna dalam mengamankan metadata citra digital. Hal ini disebabkan antara citra digital asli, citra digital setelah dienkripsi, dan citra digital setelah dilakukan dekripsi, memiliki nilai histogram sama, serta dapat mengamankan keseluruhan metadata. Akan tetapi dalam penelitian tersebut, ukuran citra digital mengalami penyusutan sebesar-25.15%, sehingga hanya dapat digunakan sebagai perlindungan hak cipta citra digital saja. Untuk dapat digunakan sebagai perlindungan barang bukti digital, nilai hashing citra asli dan citra setelah dekripsi harus sama (Wijayanto et al., 2016). Penyusutan dalam penelitian tersebut dikarenakan perbedaan tipe header metadata citra asli dengan citra setelah dekripsi, yaitu eXchangable Image File Format (EXIF) menjadi JPEG File Interchange Format (JFIF) (Wijayanto et al., 2016). Ini disebabkan

Apart from robustness and accuracy of copy-paste image forgery detection, time complexity also plays an important role to evaluate the performance of the system. In this paper, the focus point is to improve time complexity of the block-matching algorithm. Hence, a coarse-to-fine approach is applied to propose an enhanced duplicated region detection model by using sequential block clustering. Clustering minimizes the search space in block matching. This significantly improves time complexity as it eliminates several extra block-comparing operations. We determine time complexity function of the proposed algorithm to measure the performance. The experimental results and mathematical analysis demonstrate that our proposed algorithm has more improvement in time complexity when the block size is small. ª

The study of facial growth is explored in many fields of science, including anatomy, genetics, and forensics. In the field of forensics, it acts as a valuable tool for combating child pornography. The present research proposes a new method, based on relative measurements and fixed references of the human face-specifically considering measurements of the diameter of the iris (iris ratio)-for the analysis of facial growth in association with age in children and sub-adults. The experimental sample consisted of digital photographs of 1000 Brazilian subjects, aged between 6 and 22 years, distributed equally by sex and divided into five specific age groups (6, 10, 14, 18, and 22 year olds ± one month). The software package SAFF-2D® (Forensic Facial Analysis System, Brazilian Federal Police, Brazil) was used for positioning 11 landmarks on the images. Ten measurements were calculated and used as fixed references to evaluate the growth of the other measurements for each age group, as well t...

Apart from robustness and accuracy of copy-paste image forgery detection, time complexity also plays an important role to evaluate the performance of the system. In this paper, the focus point is to improve time complexity of the block-matching algorithm. Hence, a coarse-to-fine approach is applied to propose an enhanced duplicated region detection model by using sequential block clustering. Clustering minimizes the search space in block matching. This significantly improves time complexity as it eliminates several extra block-comparing operations. We determine time complexity function of the proposed algorithm to measure the performance. The experimental results and mathematical analysis demonstrate that our proposed algorithm has more improvement in time complexity when the block size is small.

We investigate if the random feature selection approach proposed in [1] to improve the robustness of forensic detectors to targeted attacks, can be extended to detectors based on deep learning features. In particular, we study the transferability of adversarial examples targeting an original CNN image manipulation detector to other detectors (a fully connected neural network and a linear SVM) that rely on a random subset of the features extracted from the flatten layer of the original network. The results we got by considering three image manipulation detection tasks (resizing, median filtering and adaptive histogram equalization), two original network architectures and three classes of attacks, show that feature randomization helps to hinder attack transferability, even if, in some cases, simply changing the architecture of the detector, or even retraining the detector is enough to prevent the transferability of the attacks.

We investigate if the random feature selection approach proposed in [1] to improve the robustness of forensic detectors to targeted attacks , can be extended to detectors based on deep learning features. In particular, we study the transferability of adversarial examples targeting an original CNN image manipulation detector to other detectors (a fully connected neural network and a linear SVM) that rely on a random subset of the features extracted from the flatten layer of the original network. The results we got by considering three image manipulation detection tasks (resizing, median filtering and adap-tive histogram equalization), two original network architectures and three classes of attacks, show that feature randomization helps to hinder attack transferability, even if, in some cases, simply changing the architecture of the detector, or even retraining the detector is enough to prevent the transferability of the attacks.

Apart from robustness and accuracy of copy–paste image forgery detection, time complexity also plays an important role to evaluate the performance of the system. In this paper, the focus point is to improve time complexity of the block-matching algorithm. Hence, a coarse-to-fine approach is applied to propose an enhanced duplicated region detection model by using sequential block clustering. Clustering minimizes the search space in block matching. This significantly improves time complexity as it eliminates several extra block-comparing operations. We determine time complexity function of the proposed algorithm to measure the performance. The experimental results and mathematical analysis demonstrate that our proposed algorithm has more improvement in time complexity when the block size is small.

Digital image tampering becomes a common information falsification trend. Copy-Move forgery is one of the tampering types that are used. Image forgery is the science of detecting image tempering whether with a previous knowledge about the source image (active) or without (passive). In this paper, we propose a method which is efficient and fast for detecting Copy-Move regions even when the copied region was undergone rotation modify in spatial domain. The proposed method accelerates blocking matching strategy by parallel comparing between blocks. Firstly, the image is divided into fixed-size overlapping blocks then features are extracted for each block. k-means clustering technique is used to cluster the blocks into different cluster. The feature vectors of each cluster blocks are lexicographically sorted by radix sort, and then a similarity measure is calculated between each nearby blocks to determine their similarity. The experimental results show that the proposed method can detect the duplicated regions efficiently even when an image was modified by jpeg compression, rotation and smoothing conditions. The proposed system reduced processing time up to 75% of other previous works.

The study of facial growth is explored in many fields of science, including anatomy, genetics, and forensics. In the field of forensics, it acts as a valuable tool for combating child pornography. The present research proposes a new method, based on relative measurements and fixed references of the human face-specifically considering measurements of the diameter of the iris (iris ratio)-for the analysis of facial growth in association with age in children and sub-adults. The experimental sample consisted of digital photographs of 1000 Brazilian subjects, aged between 6 and 22 years, distributed equally by sex and divided into five specific age groups (6, 10, 14, 18, and 22 year olds ± one month). The software package SAFF-2D® (Forensic Facial Analysis System, Brazilian Federal Police, Brazil) was used for positioning 11 landmarks on the images. Ten measurements were calculated and used as fixed references to evaluate the growth of the other measurements for each age group, as well t...

Copy move (CM) is a common type of forgery which creates tampered image by covering some important part of the image by replacing it with some other part of the same image. Therefore, forgery detection techniques are required to identify tampered areas. In this paper, a copy-move forgery detection (CMFD) method is presented by utilizing Stationary Wavelet transform (SWT) for decomposing the image. After decomposition, approximation band is further subdivided into overlapping blocks and then features are extracted by using principal component analysis (PCA) for each block. In order to match the similar block pairs and identify areas that are likely to be tampered, euclidean distance is used to compute the block distance between the adjacent pair of blocks. To evaluate the performance of the proposed work uncompressed color image dataset (UCID) is used. The quantitative results have been discussed in terms of parameters true positive rate (TPR), false positive rate (FPR) and accuracy and also compared with other existing methods. For qualitative analysis detected images obtained from the proposed method is presented. From the results obtained it can be seen that the proposed method performs better in terms of all parameters as compare to existing methods.

Image forgery is nowadays widely used as digital images are easy to manipulate due to high availability of powerful image processing tools. It is possible to add or remove objects from an image without leaving any visible traces of tampering. This paper describes a method for detection of copy-paste manipulation on JPEG digital images. It is a type of image forgery in which a part of the image is copied to another location in the image with the intent to cover or add an important image object. The detection method was implemented through extracting and analyzing blocking artifact grids (BAGs), introduced by block processing during JPEG compression. Analysis was based on fact that BAGs usually mismatch after performing copy-paste operations. Proposed method was demonstrated on two doctored images.

Postmortem analysis of the ocular globe is an important topic for forensic pathology and transplantology. Although crucial elements may be gathered from examining cadaveric eyes, the latter do not routinely undergo in-depth analysis. The paucity of quantitative and objective data that are obtainable using current, invasive necroscopic techniques is the main reason for the limited interest in this highly specialized procedure. The aim of the current study is to describe and to object for the first time the postmortem ocular changes by mean of portable optical coherence tomography for evaluating ocular tissues postmortem. The design involved the postmortem analysis (in situ, and without enucleation) of 12 eyes by portable spectral-domain Optical Coherence Tomography. The scans were performed, in corneal, retinal and angle modality at different intervals: <6 h, 6th, 12th, and 24th hour and after autopsy (25th–72nd hour). The morphological changes in the cornea, sclera, vitreous humo...

Image forgery detection is currently one of the interested research fields of image processing. Copy-Move (CM) forgery is one of the most commonly techniques. In this paper, we propose an efficient methodology for fast CM forgery detection. The proposed method accelerates blocking matching strategy. Firstly, the image is divided into fixed-size overlapping blocks then Discrete Cosine Transform (DCT) is applied to each block to represent its features, which are used to indirectly compare the blocks. After sorting the blocks based on DCT coefficients, a distance is measured between nearby blocks to denote their similarity. The proposed Fan Search (FS) algorithm starts once a duplicated block is detected. Instead of exhaustive search for all blocks, the nearby blocks of the detected block are examined first in a spiral order. The experimental results demonstrate that the proposed method can detect the duplicated regions efficiently, and reduce processing time up to 75% less than other previous works.

Digital image tampering becomes a common information falsification trend. Copy-Move forgery is one of the tampering types that are used. Image forgery is the science of detecting image tempering whether with a previous knowledge about the source image (active) or without (passive). In this paper, we propose a method which is efficient and fast for detecting Copy-Move regions even when the copied region was undergone rotation modify in spatial domain. The proposed method accelerates blocking matching strategy by parallel comparing between blocks. Firstly, the image is divided into fixed-size overlapping blocks then features are extracted for each block. k-means clustering technique is used to cluster the blocks into different cluster. The feature vectors of each cluster blocks are lexicographically sorted by radix sort, and then a similarity measure is calculated between each nearby blocks to determine their similarity. The experimental results show that the proposed method can detect the duplicated regions efficiently even when an image was modified by jpeg compression, rotation and smoothing conditions. The proposed system reduced processing time up to 75% of other previous works.

We investigate if the random feature selection approach proposed in [1] to improve the robustness of forensic detectors to targeted attacks , can be extended to detectors based on deep learning features. In particular, we study the transferability of adversarial examples targeting an original CNN image manipulation detector to other detectors (a fully connected neural network and a linear SVM) that rely on a random subset of the features extracted from the flatten layer of the original network. The results we got by considering three image manipulation detection tasks (resizing, median filtering and adap-tive histogram equalization), two original network architectures and three classes of attacks, show that feature randomization helps to hinder attack transferability, even if, in some cases, simply changing the architecture of the detector, or even retraining the detector is enough to prevent the transferability of the attacks.

Image retrieval from large databases, such as popular social networks, collections of surveillance images and videos, or digital investigation archives, is a very important task for a number of applications. In digital investigation, hashing techniques are commonly used to index large quantities of images to detect copies from different archives. In the last few years, a number of image hashing techniques based on the Bags of Visual Words paradigm have been proposed. Recently, this paradigm has been augmented by using multiple descriptors (Bags of Visual Phrases) to exploit the coherence between different feature spaces. In this paper we propose to further improve the Bags of Visual Phrases approach exploiting the coherence between feature spaces not only in the image representation, but also in the codebooks generation. Experiments performed on real and synthetic near duplicate image datasets show the effectiveness of the proposed approach, which outperforms the original Bags of Visual Phrases approach.

Digital image tampering becomes a common information falsification trend. Copy-Move forgery is one of the tampering types that are used. Image forgery is the science of detecting image tempering whether with a previous knowledge about the source image (active) or without (passive). In this paper, we propose a method which is efficient and fast for detecting Copy-Move regions even when the copied region was undergone rotation modify in spatial domain. The proposed method accelerates blocking matching strategy by parallel comparing between blocks. Firstly, the image is divided into fixed-size overlapping blocks then features are extracted for each block. k-means clustering technique is used to cluster the blocks into different cluster. The feature vectors of each cluster blocks are lexicographically sorted by radix sort, and then a similarity measure is calculated between each nearby blocks to determine their similarity. The experimental results show that the proposed method can detect the duplicated regions efficiently even when an image was modified by jpeg compression, rotation and smoothing conditions. The proposed system reduced processing time up to 75% of other previous works.

Whilst it is sometimes essential that a scene is well lit before image capture, too much light can cause exposure or glare-based problems. Typically, glare is introduced to images when the camera is pointed towards the light source, and results in a visible distortion in the image. In this paper, we analyse and identify images that contain the 'glare' property using the empirical Benford's Law. The experiment is performed on 1338 images, and extracts discrete wavelet High High (HH), High Low (HL) and Low High (LH) sub bands as raw data. The significant digit from each coefficient of all sub bands is then calculated. We then analyse the probability of occurrence of large digits against smaller digits to detect anomalies. All images containing these anomalies are further analysed for the identification of additional salient features. This analysis is performed in accordance with the Benford's Law plot and the help of probability intensity histogram and divergence. Our results indicate that 142 images have irregular Benford's Law curves. For most images, the irregularity occurs at the 5th digit. After visual examination, we have found the unbalanced light and high level of brightness in these images. To measure the intensity of light in an image, we compute the probability histogram of gray levels. These results also correlate with the irregular peak identified from the Benford's Law curves. In addition, the divergence is then computed, which shows the deviation between the actual Benford's Law curve and the Benford's Law graph of an image. Our proposed technique is novel and has a potential to be an image forensic tool for quick image analysis.

The distribution of digital images with the classic and newest technologies available on Internet (e.g., emails, social networks, digital repositories) has induced a growing interest on systems able to protect the visual content against malicious manipulations that could be performed during their transmission. One of the main problems addressed in this context is the authentication of the image received in a communication. This task is usually performed by localizing the regions of the image which have been tampered. To this aim the received image should be first registered with the one at the sender by exploiting the information provided by a specific component of the forensic hash associated with the image. In this paper we propose a robust alignment method which makes use of an image hash component based on the Bag of Visual Words paradigm. The proposed signature is attached to the image before transmission and then analyzed at destination to recover the geometric transformations which have been applied to the received image. The estimator is based on a voting procedure in the parameter space of the geometric model used to recover the transformation occurred to the received image. Experiments show that the proposed approach obtains good margin in terms of performances with respect to state-of-the art methods.