A survey of machine learning techniques in adversarial image forensics

Image forgeries can be detected and localized by using deep convolution neural network, and semantic segmentation. Color illumination is used to apply color map after pre-processing step. To train VGG-16 with two classes using deep convolution neural network transfer learning approach is used. This algorithm classifies image’s pixels having a forgery or not. These classified images with color pixel label are trained using semantic segmentation to localize forged pixels. These algorithms are tested on GRIP, DVMM, CMFD, and BSDS300 datasets. All these images are divided into two folders. One folder contains all forged images, and another folder contains labels of forged pixels. The experiment result shows that total accuracy is 0.98482, average accuracy is 0.98581, average IoU is 0.91148, weighted IoU is 0.97193, and average boundary F1 score is 0.86404. The forged pixel accuracy is 0.98698, IoU of the forged pixel is 0.83945, and average boundary F1 score of the forged image is 0.79709. Not Forged pixel accuracy is 0.98463, IoU of not forged pixel is 0.98351 and average boundary F1 score of not forged image is 0.93055. The experiment results show that forged pixel and not forged detection accuracy is above 98%, which is best among other methods.

Copy move and splicing forgery detection using deep convolution neural network, and semantic segmentation

Fractional Fourier Transform based Riesz fractional derivative approach for edge detection and its application in image enhancement

Image forensic plays a crucial role in both criminal investigations (e.g., dissemination of fake images to spread racial hate or false narratives about specific ethnicity groups) and civil litigation (e.g., defamation). Increasingly, machine learning approaches are also utilized in image forensics. However, there are also a number of limitations and vulnerabilities associated with machine learning-based approaches, for example how to detect adversarial (image) examples, with real-world consequences (e.g., inadmissible evidence, or wrongful conviction). Therefore, with a focus on image forensics, this paper surveys techniques that can be used to enhance the robustness of machine learning-based binary manipulation detectors in various adversarial scenarios.

A Survey of Machine Learning Techniques in Adversarial Image Forensics

An efficient approach for forgery detection in digital images using Hilbert–Huang transform

Improved JPEG anti-forensics with better image visual quality and forensic undetectability.

With the increase in demand for identification of authenticity of the digital images, researchers are widely studying the image forgery detection techniques. Copy-move forgery is amongst the commonly used forgery, which is performed by copying a part of an image and then pasting it on the same or different image. This results in the concealing of image content. Most of the existing copy-move forgery detection techniques are subjected to degradation in results, under the effect of geometric transformations. In this paper, a Discrete Cosine Transformation (DCT) and Singular Value Decomposition (SVD) based technique is proposed to detect the copy-move image forgery. DCT is used to transform the image from the spatial domain to the frequency domain and SVD is used to reduce the feature vector dimension. Combination of DCT and SVD makes the proposed scheme robust against compression, geometric transformations, and noise. For classification of images as forged or authentic, Support Vector Machine (SVM) classifier is used on the feature set. Once the image is detected as forged, then for the localization of forged region, K-means clustering is used on the feature vector. According to the distance threshold, similar blocks are identified and marked. The application of SVD provides stability and invariance from geometric transformations. Evaluation of the proposed scheme is done with and without post-processing operations on the images, both at the pixel level and image level. The proposed scheme outperforms the various state-of-the-art techniques of Copy-Move Forgery Detection (CMFD) in terms of accuracy, precision, recall and F1 parameters. Moreover, the proposed scheme also provides better results against rotation, scaling, noise and JPEG compression.

An improved block based copy-move forgery detection technique

The forensic investigation of JPEG compression generally relies on the analysis of first-order statistics based on image histogram. The JPEG compression detection methods based on such methodology can be effortlessly circumvented by adopting some anti-forensic attacks. This paper presents a counter JPEG anti-forensic method by considering the second-order statistical analysis based on the co-occurrence matrices (CMs). The proposed framework comprises three stages: selection of the target difference image, evaluation of CMs, and generation of second-order statistical feature based on CMs. In the first stage, we explore the effects of dithering operation of JPEG anti-forensics by analyzing the variance inconsistencies along the diagonals. Afterward, CMs are evaluated in the second stage to highlight the effects of grainy noise introduced during the dithering operation. The third stage is devoted to generate an optimal second-order statistical feature which is fed to the SVM classifier. The experimental results based on the uncompressed color image database and BOSSBase dataset images demonstrated that the proposed forensic detector based on CM is very efficient even in the presence of anti-forensic attacks. Moreover, the experimental results also confirm the competency of the proposed method in counter median filtering and contrast enhancement anti-forensics. The proposed scheme also provides satisfactory results in detecting other image processing operations such as mean filtering, Gaussian filtering, Weiner filtering, scaling, and rotation, thereby revealing its multi-purpose nature.

Counter JPEG Anti-Forensic Approach Based on the Second-Order Statistical Analysis

The image acquisition device, the light is filtered through a Color Filter Array (CFA), where each pixel captures only one color (from Red, Green, and Blue), while others are calibrated. This process is known as interpolation process, and the artifacts introduced are called CFA or interpolation artifacts. The structure of these artifacts in the image is disturbed while a forgery is introduced in an image. In this paper, a high-order statistical approach is proposed to detect the inconsistencies in the artifacts of different parts of the image to expose any forgery present. The Markov Transition Probability Matrix (MTPM) is employed to develop various features that will detect the presence or absence of CFA artifacts in a particular region of the image. The Markov random process is applied because it provides an enhanced efficiency and reduced computational complexity for the forgery detection model. The algorithm is tested on 2 × 2 pixel block of the image which provides the results of a fine quality. There is no prior information of the location of the forged region of the image. The algorithm is tested on various images, taken from various social networking websites. The proposed forgery detection technique outperforms the existing state-of-the-art techniques for the different forgery scenarios by providing an average accuracy of 90.58%.

A Markov based image forgery detection approach by analyzing CFA artifacts

Median filtering has received considerable attention and popularity for image enhancement and anti-forensics. It can be utilized as an image denoising and smoothing tool to disguise the footprints of image processing operations such as image resampling and JPEG compression. A two-step median filtering anti-forensic framework is proposed in this paper to fool the existing median filtering forensic detectors by hiding the median filtering artifacts. In the proposed framework, a variational deconvolution approach is initially employed to generate a median filtered forgery. Now, this forgery is further processed in the second step by considering Total Variation (TV) based minimization optimization problem to eradicate the median filtering artifacts left during deconvolution operation. Moreover, the proposed TV-based minimization algorithms significantly reduce the unnatural (grainy) noise left during the variational deconvolution. Two types of TV-based minimization problems are suggested, first relies on the TV of energy by considering the image gradient and second on the structure of a given image. The performance of the proposed scheme is evaluated by considering the worst-case and optimal scenarios. The experimental results based on UCID and BOSSBase dataset images demonstrate that the proposed anti-forensic methods provide superior results in terms of image visual quality and forensic undetectability as compared to the existing approaches, with slight increase in computational time.

Gurinder Singh

Papers

Improved JPEG anti-forensics with better image visual quality and forensic undetectability.

An improved block based copy-move forgery detection technique

Counter JPEG Anti-Forensic Approach Based on the Second-Order Statistical Analysis

A Markov based image forgery detection approach by analyzing CFA artifacts

An improved median filtering anti-forensics with better image quality and forensic undetectability