Manipulation Classification for JPEG Images Using Multi-Domain Features

Abstract

Image forensics comprises the analyses and classifications of manipulations that have been applied to images. The ability to classify various manipulations that have been employed in the process of forgery is essential. Techniques to identify multiple manipulations applied to uncompressed images have been reported thus far, but the forensic approach for JPEG images compressed with various qualities has not been proposed. In this paper, we propose the manipulation classification network (MCNet) to exploit multi-domain features of the spatial, frequency, and compression domains. The proposed MCNet learns several forensic features for each domain through a multi-stream structure and distinguishes manipulations by comprehensively analyzing the fused features. Our work jointly considers visual artifacts caused by image manipulations and compression artifacts due to JPEG compression; therefore, rich forensic features can be explored and learned in the training phase. To enable forgery analysis in the real-world environment, data were generated based on twenty types of manipulation algorithms and various compression parameters. To demonstrate the effectiveness of the proposed MCNet, extensive experiments were conducted using state-of-the-art baselines. Compared to these baselines, our proposed method outperforms in terms of multiclass manipulation classification. In addition, we experimentally proved that the fine-tuned model based on the multi-class manipulation task was effective for different forensic tasks such as DeepFake detection or integrity authentication of JPEG images.

Interpol review of forensic video analysis, 2019–2022

“…[ 53 ] proposes a method to classify the various types of manipulations that have occurred on images. The manipulation classification network (MCNet) exploits multi-domain features to extract multiple features from the frequency, spatial, and compression domains [ 53 ].…”

An Overview of Recent Work in Media Forensics: Methods and Threats

“…Note that the manipulated portion of the image could could be synthesized. Some common examples of image manipulation techniques include splicing [41] (replacing a section of an image with a section from another image), inpainting [42] (deleting a section of an image and synthesizing pixels to replace the content), copy-move [43] (duplicating a section of an image and moving it to another position within the same image), and photo-montage [44] (composite image from a combination of two or more images). Figure 1 shows examples of splicing and copy-move manipulations.…”

“…Zhu et al decomposed face images into their 3D geometry and lighting parameters [74]. Other methods use media information, such as JPEG compression artifacts, for manipulation detection and localization [43,75,76]. Bonettini et al [77] proposed a manipulation detection method that focuses on the absence of camera sensor noise in images.…”

Frame-rate Up-conversion Detection Based on Convolutional Neural Network for Learning Spatiotemporal Features

“…In the early stages of the study, many researchers analyzed of neural network-based methods for detecting image forgery. And the scope of the research varies as follows: image manipulation classification [5,6,12,29,53], double JPEG detection [3,4,35], computer graphics detection [17,52], content-aware retargeting detection [32,33], and camera model identification [10,44]. Compared with conventional methods, the network-based approaches have achieved excellent performance; hence, their interest within CNN-based forensics is steadily increasing.…”