Image Authentication Research Articles

The War in Ukraine Through the Prism of Visual Disinformation and the Limits of Specialized Fact-Checking. A Case-Study at Le Monde

The Russian invasion of Ukraine in February 2022 led to a flood of disinformation circulating online, making fact-checking and verification crucial for journalists covering the conflict. Based on five semi-structured interviews and an analysis of 48 debunks published between February and October 2022, this study aims to understand how Les Décodeurs, the fact-checking unit of the French daily newspaper Le Monde, organized its practices in response to the flow of visual disinformation surrounding the conflict, and to place them within the broader framework of practices and categorizations of previous research on fact-checking and disinformation. The findings show that most fact-checks focuses on authentic images with misleading captions, rather than sophisticated manipulation. Previous categorizations prove their ability to describe the 48 pieces of visual disinformation but a more refined approach is suggested by considering visual disinformation as mainly multi-modal. The study reveals the fact-checkers’ role as gatekeepers in filtering content from social networks and addressing the challenge of countering viral disinformation while avoiding unintended amplification. It also identifies a blend of shared and individualized approaches among fact-checking journalists, including manual and pre-digital methods, reflecting the adaptable nature of their work. Based on the latest developments of gatekeeping functions, the study highlights limitations in the challenge of reconciling reactive and retroactive practices with mid-term editorial considerations in a rapidly evolving media ecosystem.

Robust residual convolutional neural network based pupil tracking for low-computational power applications

Existing Deep Fully Convolutional Network (DFCN) based pupil segmentation models have been shown to perform robust and accurate pupil detection. However, they lack high computational resources when real-time predictions for low computational wearable devices such as Raspberry Pi are of interest. Simple FCN (Fully Convolutional Network) models, on the other hand, can provide real-time yet not as accurate predictions with low computational power devices. In this study, we address this dilemma by proposing a CNN (Convolutional Neural Network)-based model, Residual CNN, with several advanced operations such as residual connections, Squeeze and Excitation (SE) attention, and Atrous Spatial Pyramid Pooling (ASPP) to improve the prediction performance without complicating the model. Moreover, we leverage transfer learning by training synthetic images, then fine-tuning them with authentic eye images. We also fully quantize our model parameters to speed up the predictions and apply the Quantization Aware Training (QAT) strategy to obtain accurate predictions. Our experiments show that our quantized Res-CNN model trained by QAT strategy with 40x30 resolution images provides robust yet real-time predictions with an average of 1.351 Root Mean Square Error (RMSE) for the pupil center predictions and 8.317 ms response time per image on Raspberry Pi.

A distortion-free authentication method for color images with tampering localization and self-recovery

For immediate access and online backup, cloud storage has become a mainstream way for digital data storage and distribution. To assure images accessed or downloaded from clouds are reliable is critical to storage service providers. In this study, a new distortion-free color image authentication method based on secret sharing, data compression and image interpolation with the tampering recovery capability is proposed. The proposed method generates elaborate authentication signals which have double functions of tampering localization and image repairing. The authentication signals are subsequently converted into many shares with the use of (k, n)-threshold method so as to increase the multiplicity of authentication signals for reinforcing the capability of tampering recovery. These shares are then randomly concealed in the alpha channel of the to-be-protected image that has been transformed into the PNG format containing RGBA channels. In authentication, the authentication signals computed from the alpha channel are not only used for indicating if an image block has been tampered with or not, but used as a signal to find the corresponding color in a predefined palette to recover the tampered image block. Compared with several state-of-the-art methods, the proposed method attains positive properties including losslessness, tampering localization and tampering recovery. Experimental results and discussions on security consideration and comparison with other related methods are provided to demonstrate the outperformance of the proposed method.

Three-dimensional image authentication from multi-view images.

Three-dimensional (3D) optical authentication is important for modern information security. Existing 3D optical authentication methods rely on integral imaging devices, necessitating meticulous calibration and incurring high transmission overhead. To streamline the acquisition of 3D information, this paper introduces a novel 3D optical authentication approach, to the best of our knowledge, based on the construction of 3D data from multi-view images. The proposed method simplifies 3D projection by generating fixed-viewpoint elemental images, eliminating the need for additional viewpoint information during transmission and authentication. Compressed sensing is used for compression during transmission, and a deep learning network is designed for 3D reconstruction, enhancing the recovery. Experimental outcomes confirm the efficiency of our proposed approach for 3D authentication across diverse datasets.

Color-Transfer-Enhanced Data Construction and Validation for Deep Learning-Based Upper Gastrointestinal Landmark Classification in Wireless Capsule Endoscopy.

While the adoption of wireless capsule endoscopy (WCE) has been steadily increasing, its primary application remains limited to observing the small intestine, with relatively less application in the upper gastrointestinal tract. However, there is a growing anticipation that advancements in capsule endoscopy technology will lead to a significant increase in its application in upper gastrointestinal examinations. This study addresses the underexplored domain of landmark identification within the upper gastrointestinal tract using WCE, acknowledging the limited research and public datasets available in this emerging field. To contribute to the future development of WCE for gastroscopy, a novel approach is proposed. Utilizing color transfer techniques, a simulated WCE dataset tailored for the upper gastrointestinal tract is created. Using Euclidean distance measurements, the similarity between this color-transferred dataset and authentic WCE images is verified. Pioneering the exploration of anatomical landmark classification with WCE data, this study integrates similarity evaluation with image preprocessing and deep learning techniques, specifically employing the DenseNet169 model. As a result, utilizing the color-transferred dataset achieves an anatomical landmark classification accuracy exceeding 90% in the upper gastrointestinal tract. Furthermore, the application of sharpen and detail filters demonstrates an increase in classification accuracy from 91.32% to 94.06%.

Privacy-Preserving Image Storage on Cloud Using An Unified Cryptographic Authentication Scheme

With the proliferation of several cutting-edge technologies such as the Artificial Intelligence (AI), and Machine Learning (ML), Internet of Things (IoT), cloud technology is gaining colossal popularity in recent years. Despite the general publicity on the theme across the digital world, defending user data kept in the cloud database is the most decisive problem. Recent potential cyber attacks reveal that storing private images entails more unique care related to other types of information on the cloud. As the cloud customer who has kept their images has no control over their data the cloud service provider has to ensure better security against cyber threats. Cryptography algorithms are the best choice to secure pictorial data in the cloud. These techniques transform images into an inarticulate form to keep confidentiality over undependable and vulnerable social media .In this paper, we aim to propose an approach for improving image security on the cloud using cryptography algorithms. We developed a cohesive approach, called Unified Cryptographic Image Authentication (UCIA) to protect user images on a cloud platform. The proposed UCIA approach includes two phases: (i)UCIA engenders a cipher text through a Data Encryption Standard (DES) by providing a key and a message as input, and (ii)UCIA implements a Twofish algorithm to encipher the pictures by applying cipher text. The enciphered picture data is then stored in the cloud database and can be recovered when the customer requests it. The effectiveness of both enciphering and deciphering procedures are analyzed using the evaluation metrics including time for enciphering, deciphering, cloud storage, and enciphering throughput. Experimental results reveal the better performance and strength of the UCIA approach

NAISS: A reverse proxy approach to mitigate MageCart's e-skimmers in e-commerce

The rise of payment details theft has led to increasing concerns regarding the security of e-commerce platforms. For the MageCart threat family, the attacks employ e-skimmers, which are pieces of software code that instruct clients to forward payment details to an attacker-controlled server. They can be injected into hosting providers' servers as HTML tags such as script, iframe, and img. By leveraging image steganography - the technique of hiding structured information inside images without visual perturbances - MageCart groups can deliver e-skimmers without raising suspicion. In this work, we systematically review applicable solutions in the literature and evaluate their drawbacks in the setting of a compromised hosting provider. While promising, existing solutions in the literature present shortcomings such as a lack of compatibility, adaptability, or functionality in the presence of an attacker. Based on this review, we compile a set of features for a better solution, which we use as a foundation for designing our proposed solution - NAISS: Network Authentication of Images to Stop e-Skimmers. Through our solution, digital signatures of individual images are checked inside a server-side middlebox residing in the hosting provider's network to prevent the transmission of unauthorized images to clients. Elliptic curve signatures are provided by the e-commerce platform developer prior to uploading a website to the hosting provider. Our proof-of-concept implementation shows that NAISS is capable of filtering 100% of present stegoimages, regardless of their novelty, while imposing a minimal performance detriment and no client-side modifications.

Blockchain based medical image encryption using Arnold’s cat map in a cloud environment

Improved software for processing medical images has inspired tremendous interest in modern medicine in recent years. Modern healthcare equipment generates huge amounts of data, such as scanned medical images and computerized patient information, which must be secured for future use. Diversity in the healthcare industry, namely in the form of medical data, is one of the largest challenges for researchers. Cloud environment and the Block chain technology have both demonstrated their own use. The purpose of this study is to combine both technologies for safe and secure transaction. Storing or sending medical data through public clouds exposes information into potential eavesdropping, data breaches and unauthorized access. Encrypting data before transmission is crucial to mitigate these security risks. As a result, a Blockchain based Chaotic Arnold’s cat map Encryption Scheme (BCAES) is proposed in this paper. The BCAES first encrypts the image using Arnold’s cat map encryption scheme and then sends the encrypted image into Cloud Server and stores the signed document of plain image into blockchain. As blockchain is often considered more secure due to its distributed nature and consensus mechanism, data receiver will ensure data integrity and authenticity of image after decryption using signed document stored into the blockchain. Various analysis techniques have been used to examine the proposed scheme. The results of analysis like key sensitivity analysis, key space analysis, Information Entropy, histogram correlation of adjacent pixels, Number of Pixel Change Rate, Peak Signal Noise Ratio, Unified Average Changing Intensity, and similarity analysis like Mean Square Error, and Structural Similarity Index Measure illustrated that our proposed scheme is an efficient encryption scheme as compared to some recent literature. Our current achievements surpass all previous endeavors, setting a new standard of excellence.

Color Patterns And Enhanced Texture Learning For Detecting Computer-Generated Images

Abstract Detection of computer-generated (CG) images can reveal the authenticity and originality of digital images. However, recent cutting-edge image generation methods make it very difficult to distinguish CG images from natural photographs. In this paper, a novel method based on color patterns and enhanced texture learning is proposed to tackle this problem. We designed and implemented the backbone network with a separation-fusion learning strategy by constructing a multi-branch neural network. The luminance and chrominance patterns in dual-color spaces (RGB and YCbCr) are leveraged to achieve a robust representation of image differences. A channel-spatial attention module and a global texture enhancement module are also integrated into a backbone network to enhance the learning of inherent traces. Experiments on several commonly used benchmark datasets and a newly constructed dataset with more realistic and diverse images demonstrate that the proposed algorithm outperforms state-of-the-art competitors by a large margin.

Privacy-Preserving Image Storage on Cloud Using An Unified Cryptographic Authentication Scheme

With the proliferation of several cutting-edge technologies such as the Artificial Intelligence (AI), and Machine Learning (ML), Internet of Things (IoT), cloud technology is gaining colossal popularity in recent years. Despite the general publicity on the theme across the digital world, defending user data kept in the cloud database is the most decisive problem. Recent potential cyber attacks reveal that storing private images entails more unique care related to other types of information on the cloud. As the cloud customer who has kept their images has no control over their data the cloud service provider has to ensure better security against cyber threats. Cryptography algorithms are the best choice to secure pictorial data in the cloud. These techniques transform images into an inarticulate form to keep confidentiality over undependable and vulnerable social media .In this paper, we aim to propose an approach for improving image security on the cloud using cryptography algorithms. We developed a cohesive approach, called Unified Cryptographic Image Authentication (UCIA) to protect user images on a cloud platform. The proposed UCIA approach includes two phases: (i)UCIA engenders a cipher text through a Data Encryption Standard (DES) by providing a key and a message as input, and (ii)UCIA implements a Twofish algorithm to encipher the pictures by applying cipher text. The enciphered picture data is then stored in the cloud database and can be recovered when the customer requests it. The effectiveness of both enciphering and deciphering procedures are analyzed using the evaluation metrics including time for enciphering, deciphering, cloud storage, and enciphering throughput. Experimental results reveal the better performance and strength of the UCIA approach.

Detecting facial image forgeries with transfer learning techniques

Digital images have become ubiquitous in our daily lives, appearing on our smartphone screens and online websites. They are widely used in numerous industries, including media, forensic and criminal investigations, medicine, and more. The ease of access to consumer photo editing tools has made it simple to manipulate images. However, such altered images pose a serious risk in fields where image authenticity is crucial, making it challenging to confirm the reliability of digital images. Digital image fraud involves altering an image's meaning without leaving any obvious signs. In this study, we present three convolutional neural network-based transfer learning techniques “CNN” classification of facial image forgeries, using VGG-19, InceptionV3, and DenseNet201. Among these methods, DenseNet201 achieved the highest accuracy of 99%, followed by InceptionV3 at 94% and VGG-19 at 84%.

A novel minority sample fault diagnosis method based on multisource data enhancement

AbstractEffective fault diagnosis has a crucial impact on the safety and cost of complex manufacturing systems. However, the complex structure of the collected multisource data and scarcity of fault samples make it difficult to accurately identify multiple fault conditions. To address this challenge, this paper proposes a novel deep‐learning model for multisource data augmentation and small sample fault diagnosis. The raw multisource data are first converted into two‐dimensional images using the Gramian Angular Field, and a generator is built to transform random noise into images through transposed convolution operations. Then, two discriminators are constructed to evaluate the authenticity of input images and the fault diagnosis ability. The Vision Transformer network is built to diagnose faults and obtain the classification error for the discriminator. Furthermore, a global optimization strategy is designed to upgrade parameters in the model. The discriminators and generator compete with each other until Nash equilibrium is achieved. A real‐world multistep forging machine is adopted to compare and validate the performance of different methods. The experimental results indicate that the proposed method has multisource data augmentation and minority sample fault diagnosis capabilities. Compared with other state‐of‐the‐art models, the proposed approach has better fault diagnosis accuracy in various scenarios.

Enhanced Counterfeit Detection of Bangladesh Currency through Convolutional Neural Networks: A Deep Learning Approach

Counterfeiting poses a significant threat to the stability of Bangladesh's currency, the Taka, necessitating advanced methods for detection and prevention. This paper presents an innovative approach to counterfeit detection using Convolutional Neural Networks (CNNs), a deep learning technology. Explicitly focused on Bangladesh's currency, this method aims to enhance the accuracy and efficiency of counterfeit detection by leveraging the power of artificial intelligence. The proposed approach involves training CNNs on a dataset of authentic and counterfeit Bangladeshi currency images, allowing the network to learn intricate features and patterns indicative of counterfeit notes. By exploiting the hierarchical structure of CNNs, the system can automatically extract discriminative features from currency images, enabling robust detection of counterfeit banknotes. The CNN-based approach offers several advantages compared to traditional methods, which often rely on manual inspection or rule-based algorithms. It can handle complex visual information more accurately and efficiently, making it well-suited for detecting subtle counterfeit features. Furthermore, the adaptability of CNNs allows for continuous learning and improvement, ensuring resilience against evolving counterfeit techniques. The efficacy of the proposed method is validated through extensive experimentation and evaluation, demonstrating its superior performance in detecting counterfeit Bangladesh currency notes. By harnessing the capabilities of deep learning, this approach not only enhances the security of Bangladesh's financial system but also serves as a scalable solution applicable to other currencies and regions facing similar challenges. In conclusion, the integration of Convolutional Neural Networks represents a significant advancement in counterfeit detection technology, offering a powerful and versatile tool for safeguarding the integrity of Bangladesh's currency and combating financial fraud on a global scale.

Assessing the visual appeal of real/AI-generated food images

A study designed to investigate the ability of individuals to differentiate between AI-generated and authentic food images, as well as the impact of disclosing this information on the consumer perception of the appeal of these images is reported. Two online experiments were conducted with real and AI-generated food images stretching across the unprocessed, processed, and ultra-processed food continuum. Study 1 was designed to assess the accuracy with which people could identify AI-generated food images while Study 2 explored how the disclosure of an image’s origin influenced the appeal of the depicted food. The participants in Study 1 found it very easy to recognize the AI-generated images, particularly in the case of ultra-processed foods. Notably, without disclosure, the AI-generated images were often preferred. At the same time, however, disclosing that a food image was genuine significantly boosted its appeal, whereas the revelation that it had been generated by AI mitigated this effect. These insights help to understand consumer psychology in the rapidly-evolving digital food marketing landscape, highlighting the nuanced effects of technological advancements in AI image-generation on human perception.

A YOLOV3-Based Method for Detecting Deepfake Manipulated Facial Images

With the advancement of technology and the development of applications that make it easier to transfer images, sounds and videos to the virtual environment, it has become much easier to access people's personal information, videos and images. Deepfake technology produces fakes of authentic images or sounds using deep learning and artificial intelligence techniques. Today, in addition to being used in the entertainment and film industries, it is also used in situations such as creating fake news and discrediting people. Different studies have been conducted in the literature to detect deepfake images and videos to prevent these situations. In this study, a comprehensive literature review was conducted. Real and fake images were collected and labelled from different datasets or videos, and a dataset was created by applying the necessary pre-processing steps. With the created dataset, training was carried out with YOLOv3 technology, which calculates class probabilities differently from traditional methods using Convolutional Neural Networks (CNN) and handles all operations in a single regression problem, which can make fast and high-accurate detection, and the modelling process is explained. With the tests performed in the study, the model that can detect fake images produced with deepfake technology with 95% accuracy was obtained.

AI-assisted deepfake detection using adaptive blind image watermarking

This paper proposes a new adaptive blind watermarking technology for deepfake detection, which can embed deepfake detection information into the image and verify the image's authenticity without requiring additional information. The proposed scheme utilizes mixed modulation combined with partly sign-altered mean value to embed a set of coefficients that enhance robustness against attacks while maintaining high image quality. Additionally, blind adaptive deepfake detection technology with the tamper detection mean value is employed to detect relative positions adaptively, even when face images are slightly modified or deepfaked. To further improve the performance of the proposed scheme, a gray wolf optimizer is introduced to optimize parameters, and a denoising autoencoder is employed to facilitate the identification of extracted watermarks. This technology will adaptively embed watermark information while preserving the original face image, thereby maintaining the authenticity of the face in the image and verifying the owner of the image. The code is available at https://github.com/lyhsu01/AwDD.

Synthetic CT imaging for PET monitoring in proton therapy: a simulation study.

Objective.This study addresses a fundamental limitation of in-beam positron emission tomography (IB-PET) in proton therapy: the lack of direct anatomical representation in the images it produces. We aim to overcome this shortcoming by pioneering the application of deep learning techniques to create synthetic control CT images (sCT) from combining IB-PET and planning CT scan data.Approach.We conducted simulations involving six patients who underwent irradiation with proton beams. Leveraging the architecture of a visual transformer (ViT) neural network, we developed a model to generate sCT images of these patients using the planning CT scans and the inter-fractional simulated PET activity maps during irradiation. To evaluate the model's performance, a comparison was conducted between the sCT images produced by the ViT model and the authentic control CT images-serving as the benchmark.Main results.The structural similarity index was computed at a mean value across all patients of 0.91, while the mean absolute error measured 22 Hounsfield Units (HU). Root mean squared error and peak signal-to-noise ratio values were 56 HU and 30 dB, respectively. The Dice similarity coefficient exhibited a value of 0.98. These values are comparable to or exceed those found in the literature. More than 70% of the synthetic morphological changes were found to be geometrically compatible with the ones reported in the real control CT scan.Significance.Our study presents an innovative approach to surface the hidden anatomical information of IB-PET in proton therapy. Our ViT-based model successfully generates sCT images from inter-fractional PET data and planning CT scans. Our model's performance stands on par with existing models relying on input from cone beam CT or magnetic resonance imaging, which contain more anatomical information than activity maps.

Exploring weakly-supervised image manipulation localization with tampering Edge-based class activation map

In recent investigations concerning manipulated image localization, prevailing methodologies have relied on Convolutional Neural Networks (CNNs) to facilitate supervised learning predicated upon manipulated images and their corresponding ground-truth annotations. Nevertheless, these methods produce high false detection rates when detecting unaltered images due to the limited availability of authentic images. In this paper, we initiate our inquiry by scrutinizing the distribution characteristics of class activation maps within the context of image manipulation detection and the inherent patterns of alterations within this domain. We then introduce the novel concept of a tampering Edge-based Class Activation Map (EdgeCAM) tailored for detecting tampering within manipulated images. Leveraging EdgeCAM, we formulate a framework for localizing tampered regions in images through a weakly-supervised approach, relying solely on image-level annotations. Ultimately, we establish a fundamental benchmark for weakly-supervised image manipulation localization by utilizing genuine tampered images collected from real-world scenarios. The empirical results highlight the notable efficacy of our proposed method, outperforming current weakly-supervised methods for detecting image manipulation by a significant margin of 12.7% on the average combined F1 metrics across various generic datasets. This result also achieves performance comparable to supervised methods for image manipulation localization.

A reversible fragile watermarking technique using fourier transform and Fibonacci Q-matrix for medical image authentication

Image authentication techniques are crucial for a multitude of multimedia applications. When images are transmitted through non-secure channels like the internet, they are vulnerable to unauthorized alterations and manipulations. Therefore, it is essential in sensitive fields, like the medical industry, to employ image authentication techniques to guarantee the integrity and authenticity of transmitted images. In this paper, to address these security issues and to ensure the authenticity of medical images, we present a reversible fragile watermarking model where the watermark is first generated from the cover image using DCT and then encrypted by the Fibonacci Q-matrix technique, enhancing the model’s security. Subsequently, DFT is performed on the host image and the obtained sub-band is segmented into 2 × 2 blocks, the watermark is then embedded within the frequency coefficients using a new embedding technique. Experimental results indicates that the presented model produces a great watermarked image quality, with a PSNR superior to 117 dB, while maintaining an acceptable embedding capacity of 0.25 BPP and high sensitivity to various attacks, as it was tested against 16 different attacks and could detect tampering in all of them. Furthermore, the presented scheme offers nearly perfect reversibility with a PSNR of 324 dB, which is very useful in the medical field where we need images with as little distortion as possible.

Finger Vein Identification Based on Large Kernel Convolution and Attention Mechanism.

FV (finger vein) identification is a biometric identification technology that extracts the features of FV images for identity authentication. To address the limitations of CNN-based FV identification, particularly the challenge of small receptive fields and difficulty in capturing long-range dependencies, an FV identification method named Let-Net (large kernel and attention mechanism network) was introduced, which combines local and global information. Firstly, Let-Net employs large kernels to capture a broader spectrum of spatial contextual information, utilizing deep convolution in conjunction with residual connections to curtail the volume of model parameters. Subsequently, an integrated attention mechanism is applied to augment information flow within the channel and spatial dimensions, effectively modeling global information for the extraction of crucial FV features. The experimental results on nine public datasets show that Let-Net has excellent identification performance, and the EER and accuracy rate on the FV_USM dataset can reach 0.04% and 99.77%. The parameter number and FLOPs of Let-Net are only 0.89M and 0.25G, which means that the time cost of training and reasoning of the model is low, and it is easier to deploy and integrate into various applications.