MUCKE aims to mine a large volume of images, to structure them conceptually and to use this conceptual structuring in order to improve large-scale image retrieval. The last decade witnessed important progress concerning low-level image representations. However, there are a number problems which need to be solved in order to unleash the full potential of image mining in applications. The central problem with low-level representations is the mismatch between them and the human interpretation of image content. This problem can be instantiated, for instance, by the incapability of existing descriptors to capture spatial relationships between the concepts represented or by their incapability to convey an explanation of why two images are similar in a content-based image retrieval framework. We start by assessing existing local descriptors for image classification and by proposing to use co-occurrence matrices to better capture spatial relationships in images. The main focus in MUCKE is on cleaning large scale Web image corpora and on proposing image representations which are closer to the human interpretation of images. Consequently, we introduce methods which tackle these two problems and compare results to state of the art methods. Note: some aspects of this deliverable are withheld at this time as they are pending review. Please contact the authors for a preview.

http://ieeexplore.ieee.org/iel2/4524/12820/00592460.pdf

Image Processing

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. HighlightsA new comprehensive survey of pixel-based forgery detection methods is presented.A framework for grouping different forgery detection algorithms is described.We outline the strengths and weaknesses of research efforts in forgery detection.Numerous tables and figures, analyzing existing algorithms, are discussed.An extensive list of references covering the work of the last decade is provided.

A bibliography of pixel-based blind image forgery detection techniques

With the proliferation of smartphones and social media, journalistic practices are increasingly dependent on information and images contributed by local bystanders through Internet-based applications and platforms. Verifying the images produced by these sources is integral to forming accurate news reports, given that there is very little or no control over the type of user-contributed content, and hence, images found on the Web are always likely to be the result of image tampering. In particular, image splicing, i.e. the process of taking an area from one image and placing it in another is a typical such tampering practice, often used with the goal of misinforming or manipulating Internet users. Currently, the localization of splicing traces in images found on the Web is a challenging task. In this work, we present the first, to our knowledge, exhaustive evaluation of today's state-of-the-art algorithms for splicing localization, that is, algorithms attempting to detect which pixels in an image have been tampered with as the result of such a forgery. As our aim is the application of splicing localization on images found on the Web and social media environments, we evaluate a large number of algorithms aimed at this problem on datasets that match this use case, while also evaluating algorithm robustness in the face of image degradation due to JPEG recompressions. We then extend our evaluations to a large dataset we formed by collecting real-world forgeries that have circulated the Web during the past years. We review the performance of the implemented algorithms and attempt to draw broader conclusions with respect to the robustness of splicing localization algorithms for application in Web environments, their current weaknesses, and the future of the field. Finally, we openly share the framework and the corresponding algorithm implementations to allow for further evaluations and experimentation.

Large-scale evaluation of splicing localization algorithms for web images

A critical survey of state-of-the-art image inpainting quality assessment metrics

The detection of forgeries in color images is a very important topic in forensic science Copy–move (or copy–paste) forgery is the most common form of tampering associated with color images Conventional copy–move forgeries detection techniques usually suffer from the problems of false positives and susceptibility to many signal processing operations It is a challenging work to design a robust copy–move forgery detection method In this paper, we present a novel block-based robust copy–move forgery detection approach using invariant quaternion exponent moments (QEMs) Firstly, original tempered color image is preprocessed with Gaussian low-pass filter, and the filtered color image is divided into overlapping circular blocks Then, the accurate and robust feature descriptor, QEMs modulus, is extracted from color image block holistically as a vector field Finally, exact Euclidean locality sensitive hashing is utilized to find rapidly the matching blocks, and the falsely matched block pairs are removed by customizing the random sample consensus with QEMs magnitudes differences Extensive experimental results show the efficacy of the newly proposed approach in detecting copy–paste forgeries under various challenging conditions, such as noise addition, lossy compression, scaling, and rotation We obtain the average forgery detection accuracy (F-measure) in excess of 96 and 88% across postprocessing operations, at image level and at pixel level, respectively

Robust copy–move forgery detection using quaternion exponent moments

With the advent of powerful image editing tools, manipulating images and changing their content is becoming a trivial task. It is now possible to add, modify, or remove important features from an image without leaving any perceptual traces of tampering. With more than several million pictures uploaded daily to the net, and the introduction of e-Government services, it is becoming important to develop robust detection methods to identify image tampering operations. To this end, image forensics techniques aim at restoring trust and acceptance in digital media by uncovering tampering methods. Such detection techniques are the focus of this paper. In particular, we provide a survey of different forging detection techniques with a focus on copy and move approaches.

A review on copy move image forgery detection techniques

Contrast enhancement is one of the most studied problems in image processing. A plethora of approaches has been proposed in the literature for image enhancement since the pioneer work of Kovasznay and Joseph in 1955 [1] and the famous contribution of Gabor in 1965 on image deblurring [2]. However, very few works have been dedicated to contrast enhancement evaluation. This is mainly due to the fact that image enhancement is primarily related to subjective aspects of human perceptual vision. The intent of this paper is to provide a first critical review to some contrast enhancement evaluation measures and propose a new one. An objective comparison of these measures on various color real images processed by some neighborhood based methods is provided. This work is considered as a first step towards the development of a unifying framework for image enhancement evaluation. This could be also used to control the side effect that may result from any image enhancement such as contrast enhancement, denoising, tone mapping and other similar image processing tools.

A critical look to some contrast enhancement evaluation measures

We propose a new framework for evaluating image enhancement methods. In this work, we focus on contrast enhancement, and use the joint probability and the mutual information derived from the image co-occurrence matrix to determine the proposed index. We show that the second order entropy and the mutual information can be used jointly for tracking and evaluating the effect of contrast enhancement. More importantly, we show that the variations of the mutual information mimic, in a consistent way, human perceptual sensitivity to visual appearance of image changes due to contrast enhancement. The proposed approach is seen as a completely new framework for evaluating image enhancement methods, validated by a number of experiments. Finally, we also show that the proposed metric can even be used to detect image enhancement thresholds at which unpleasant artifacts start to be perceived by viewers.

An information based framework for performance evaluation of image enhancement methods

The development of robust No-Reference Image Quality Assessment (NR-IQA) techniques continues to be a challenging problem. NR-IQA techniques are critical In numerous multimedia applications. Most existing techniques are distortion-specific, as they are only efficient when the type of distortion is known. In this work, we introduce a computationally efficient NR-IQA algorithm that uses basic filtering operations in spatial domain. The features are calculated using Laws' filters proven to be efficient in texture analysis. The image quality score is predicted using a simple Generalized Regression Neural Network. The proposed algorithm has low computational complexity, making it suitable for real-time applications. The performance of the proposed technique is confirmed, using the LIVE 2 image quality assessment dataset. The proposed approach is shown to provide excellent results that are robust across different distortions, and is computationally less expensive than most existing techniques.

M. Ali Qureshi

Papers

A review on copy move image forgery detection techniques

A critical look to some contrast enhancement evaluation measures

An information based framework for performance evaluation of image enhancement methods

A fast no reference image quality assessment using laws texture moments