JPEG Research Articles

Learning Compressed Artifact for JPEG Manipulation Localization Using Wide-Receptive-Field Network

JPEG image manipulation localization aims to accurately classify and locate tampered regions in JPEG images. Existing image manipulation localization schemes usually consider diverse data streams of spatial domain, e.g. noise inconsistency and local content inconsistency. They, however, easily ignore an objective scenario: data stream features of spatial domain are hard to directly apply to compressed image format, e.g., JPEG, because tampered JPEG images may contain severe re-compression inconsistency and re-compression artifacts, when they are re-compressed to JPEG format. As a result, the traditional localization schemes relying on general data streams of spatial domain may result in a large number of false detection of tampered region in JPEG images. To address the above problem, we a new JPEG image manipulation localization scheme, in which a wide-receptive-field attention network is designed to effectively learn JPEG compressed artifacts. We firstly introduce the wide-receptive-field attention mechanism to re-construct U-Net network, which can effectively capture contextual information of JPEG images and analyze tampering traces from different image regions. Furthermore, a flexible JPEG compressed artifact learning module is designed to capture the image noise caused by JPEG compression, in which the weights can be adjusted flexibly based on image quality, without the need for decompression operations on JPEG images. Our proposed method can significantly strength the differentiation capability of detection model for tampered and non-tampered regions. A series of experiments are performed over different image sets, and the results demonstrate that the proposed scheme can achieve an overall localization performance for multi-scale JPEG manipulation regions and outperform most of state-of-the-art schemes in terms of detection accuracy, generalization and robustness.

SSAT: Active Authorization Control and User’s Fingerprint Tracking Framework for DNN IP Protection

As training a high-performance deep neural network (DNN) model requires a large amount of data, powerful computing resources and expert knowledge, protecting well-trained DNN models from intellectual property (IP) infringement has raised serious concerns in recent years. Most existing methods using DNN watermarks to verify the ownership of the models after IP infringement occurs, which is reactive in the sense that they cannot prevent unauthorized users from using the model in the first place. Different from these methods, in this article, we propose an active authorization control and user’s fingerprint tracking method for the IP protection of DNN models by utilizing sample-specific backdoor attack. The proposed method inversely and multiplely exploits sample-specific trigger as the key to implement authorization control for DNN model, in which the generated triggers are imperceptible and sample-specific for clean images. Specifically, a U-Net model is used to generate backdoor instances. Then, the target model is trained on the clean images and backdoor instances, which are inversely labeled as wrong classes and correct classes, respectively. Only authorized users can use the target model normally by pre-processing the clean images through the U-Net model. Moreover, the images processed by the U-Net model will contain unique fingerprint that can be extracted to verify and track the corresponding user’s identity. This article is the first work that utilizes the sample-specific backdoor attack to implement active authorization control and user’s fingerprint management for DNN model under black-box scenarios. Extensive experimental results on ImageNet dataset and YouTube Aligned Face dataset demonstrate that the proposed method is effective in protecting the DNN model from unauthorized usage. Specifically, the protected model has a low inference accuracy (1.00%) for unauthorized users, while maintaining a normal inference accuracy (97.67%) for authorized users. Besides, the proposed method can achieve 100% fingerprint tracking success rates on both the ImageNet and YouTube Aligned Face datasets. Moreover, it is demonstrated that the proposed method is robust against fine-tuning attack, pruning attack, pruning attack with retraining, reverse-engineering attack, adaptive attack, and JPEG compression attack. The code is available at https://github.com/nuaaaisec/SSAT .

Implementation of Robust and Secure Watermarking Algorithm on FPGA using DCT

Digital watermarking deals with embedding digital content in a cover signal so that it becomes indiscernible to make it robust against different security threats. This work demonstrates the implementation of a robust and secure watermarking technique on a Field Programmable Gate Array (FPGA) using Discrete Cosine Transform (DCT). Block by block DCT of the cover image was computed and watermark pixels in each of the blocks were hidden using middle-frequency band coefficients. Security is ensured by encrypting the watermarkwith a secret key and Arnold transform. The resultant watermark image is robust to various threats like JPEG compression, scaling, cropping, and noise attacks. This watermarking approach requires a large amount of data to be processed at high speed. FPGA is the best choice for such an application. Also, it is reprogrammable and easily upgradable according to user application. In this work, a watermarking algorithm is implemented using Xilinx design tools on the Artix-7 FPGA board.

Quantum color image watermarking scheme based on quantum discrete cosine transform

Quantum watermarking is a technique that embeds specific information into a quantum carrier, used for digital copyright protection. This paper introduces a new quantum color image watermarking method that combines Quantum LOG (QLOG) edge detection with Quantum Discrete Cosine Transform (QDCT) technology. First, edge detection is used on the RGB channels of the quantum color image. This helps identify the best channel for embedding the watermark. Next, the image is decomposed using the quantum Haar wavelet transform (QHWT). The image is then processed with QDCT, selecting the mid-frequency parts for watermark embedding. Using QDCT improves the watermark’s resistance to JPEG compression. The invisibility of the watermark is verified by assessing its Peak Signal-to-Noise Ratio (PSNR) and histogram. Comparative analysis shows strong resistance to various noise attacks. Simulation results confirm that the watermarking technique maintains high visual quality and resists different adversarial attacks.

Application of the KARP RABIN Algorithm for Plagiarism Detection System in Thesis Proposal Submission in the Department of Informatics Engineering STMIK Kaputama

Plagiarism is a serious threat, especially to academic honesty, so a detection system that can analyze various types of documents is needed. This research develops a plagiarism detection system using Optical Character Recognition (OCR) to convert image text into digital text. Rabin – Karp algorithm with rolling hash and Dice Coefficient Similarity is applied to measure similarities between documents. Testing is carried out on .doc, .txt, .jpg files. As a result, the system can detect plagiarism well in clear text and image documents, but accuracy can decrease in low-quality images. In conclusion, the similarity of content, sentence structure, and format affects the degree of similarity, while OCR techniques work effectively even though they are limited to low-quality images.

Super Encryption Feal Algorithm and Base64 Algorithm Image File Security

In the rapidly advancing digital era, image file security has become a critical issue, especially with the increasing risks of data breaches and attacks on digital files. This study aims to enhance the security of image files by implementing a combination of two cryptographic algorithms: Fast Data Encipherment Algorithm- 4 (FEAL-4) and Base64. FEAL-4 is a symmetric encryption algorithm known for its high speed and processing efficiency, while Base64 is used for encoding binary data into ASCII format to ensure safer transmission. This research develops a super encryption system that integrates these two algorithms to protect the integrity and confidentiality of image files, particularly for BMP, JPEG, and PNG formats. The implementation was carried out using the Visual Basic programming language. The results of the study show that the combination of FEAL-4 and Base64 algorithms significantly enhances the security of image files, with a high success rate in the encryption and decryption processes.

Digital Signature Method for Images without Additional Files, Robust to JPEG Compression and Metadata Removal

The article discusses a method of digitally signing images that does not use metadata or additional files. The object for signature is an array of 8x8 pixel blocks with a discrete cosine transformation applied, which make up a JPEG file. Any known algorithm can be used as a signature algorithm, for example, RSA.The resulting digital signature is converted from a binary form to an image. The conversion method is based on encoding every few bits of the signature of one of the 64 basic functions of the discrete cosine transform. Next, the basic function is transformed into an image by an inverse discrete cosine transformation. The resulting signature, encoded as an image, is attached to the original image on the right to form a signed image.The digital signature is resistant to JPEG compression within adjustable limits. The following method is used to achieve compression resistance. Since the values in the blocks can change during compression, quantization is used – reducing the accuracy of the values. It is designed in such a way that when quantizing compressed and uncompressed images, the values in the blocks after quantization are becoming the same, which allows you to verify the validity of the digital signature. Quantization also includes a step of checking the parity of the received value to avoid misinterpretation of values when compression is strong.The quantization step does not apply to the part of the image containing the digital signature. This is due to the fact that this part consists of the basic functions of the discrete cosine transform, and even with strong compression, the corresponding basic function will still retain a large coefficient and will be unambiguously interpreted.

Blind image authentication using singular value decomposition with enhanced robustness by augmenting hamming code

Singular value decomposition (SVD) based image authentication has widespread applications for digital media protection including forensic and medical domains due to the high stability and robustness of singular values under small perturbations. However, most of these techniques are non-blind and have high redundancies for the extraction of secret data. The proposed scheme employs a blind extraction strategy by utilizing the intrinsic relationship of coefficient pairs of the U matrix of the host image blocks. This relationship is utilized to hide a secret binary bit after computing a threshold value in constant time. The host image underwent block-wise decomposition and hash-based randomization of blocks, followed by block-wise SVD for embedding secret bits in the U matrix. Hamming coded watermark is embedded for high precision extraction under active attacks. High imperceptibility and strong robustness are observed under attacks when tested across general purpose as well as image datasets and the system outperforms many existing works. Experimental observation shows the average PSNR between the host and marked image and average NCC between embedded and extracted watermark to be 36.98 dB and 0.999 respectively for the SIPI dataset whereas 52.7 dB and 1 respectively for the X-Ray dataset. The system exhibited high robustness under Salt-and-pepper noise (SAPN), Gaussian Noise (GAUN), Cropping (CRP), Median Filtering (MF), Histogram Equalization (HEQ), and Jpeg Compression (JPC) with average NCC values of 0.998, 1, 0.932, 0.999 and 0.967 respectively for X-Ray dataset. The time complexity O (M × N) is observed for both embedding and extraction algorithms for a host image of size M × N.

Brain–machine convergent evolution: Why finding parallels between brain and artificial systems is informative

Central nervous system neurons manifest a rich diversity of selectivity profiles—whose precise role is still poorly understood. Following the striking success of artificial networks, a major debate has emerged concerning their usefulness in explaining neuronal properties. Here we propose that finding parallels between artificial and neuronal networks is informative precisely because these systems are so different from each other. Our argument is based on an extension of the concept of convergent evolution—well established in biology—to the domain of artificial systems. Applying this concept to different areas and levels of the cortical hierarchy can be a powerful tool for elucidating the functional role of well-known cortical selectivities. Importantly, we further demonstrate that such parallels can uncover novel functionalities by showing that grid cells in the entorhinal cortex can be modeled to function as a set of basis functions in a lossy representation such as the well-known JPEG compression. Thus, contrary to common intuition, here we illustrate that finding parallels with artificial systems provides novel and informative insights, particularly in those cases that are far removed from realistic brain biology.

OODNet: A deep blind JPEG image compression deblocking network using out-of-distribution detection

JPEG is one of the most popular image compression techniques, with numerous applications ranging from medical imaging to surveillance systems. Since JPEG introduces the blocking artifacts to the decompressed visual signals, enhancing the quality of these images is of paramount importance. Recently, various deep neural networks have been proposed for JPEG image deblocking that can effectively reduce the blocking artifacts produced by the JPEG compression technique. However, most of these schemes could only handle decompressed images generated by a set of specific JPEG quality factor (QF) values employed in the network training process. Therefore, when the images are obtained by the JPEG QF values other than those used in the network training process, the performance of deep learning-based JPEG image deblocking schemes drops significantly. To address this, in this paper, we propose a novel deep learning-based blind JPEG image deblocking method, which employs out-of-distribution detection to perform deblocking efficiently for various quality factor (QF) values. The proposed scheme can distinguish between the decompressed images using the QF values used in the training set and those using the QF values not used in the training set, and then, a suitable deblocking strategy for generating high-quality images is developed. The proposed scheme is shown to outperform the state-of-the-art JPEG image deblocking methods for various QF values.

UAV forest fire detection based on CNN and InceptionV3

Abstract Forest fires have created a severe hazard in our ecological environment with abrasive climate change all over the world. What is even more concerning, the frequency of forest fires is increasing year by year. By far, a great number of methodologies, such as satellite systems, IR, and WSN, have been created to resolve this rigorous problem. However, most of these detection methods are costly and structurally complex, making them difficult to quickly learn and use, thus not effectively reducing personnel and property injuries. Our research proposed using UAVs (Unmanned Aerial Vehicles) as visual detection carriers and using InceptionV3 for visual processing of aerial images. The flame dataset from IEEE Dataport, containing 63178 JPEG pictures, is used in our research. Our proposed forest fire detection methodology using Convolutional Neural Network (CNN) and InceptionV3 achieved a remarkable 89.90% validation accuracy. Moreover, our methodology gained an ideal precision and recall ratio.

Improved JPEG Lossless Compression for Compression of Intermediate Layers in Neural Networks Based on Compute-In-Memory

With the development of Convolutional Neural Networks (CNNs), there is a growing requirement for their deployment on edge devices. At the same time, Compute-In-Memory (CIM) technology has gained significant attention in edge CNN applications due to its ability to minimize data movement between memory and computing units. However, the deployment of complex deep neural network models on edge devices with restricted hardware resources continues to be challenged by a lack of adequate storage for intermediate layer data. In this article, we propose an optimized JPEG Lossless Compression (JPEG-LS) algorithm that implements serial context parameter updating alongside parallel encoding. This method is designed for the global prediction and efficient compression of intermediate data layers in neural networks employing CIM techniques. The results indicate average compression ratios of 6.44× for VGG16, 3.62× for ResNet34, 1.67× for MobileNetV2, and 2.31× for InceptionV3. Moreover, the implementation achieves a data throughput of 32 bits per cycle at 600 MHz on the TSMC 28 nm, with a hardware cost of 122 K Gate Count.

Gender prediction using geometric morphometry with parameters of the cranium obtained from computed tomography images

Purpose: The gender difference of the cranium skeleton is of great importance in forensic anthropology and forensic medicine sciences. This study is based on this hypothesis and the gender prediction rate was obtained by processing cranium images obtained from computed tomography (CT) using geometric morphometry. Materials and Methods: CT images of 200 individuals between the ages of 25 and 65 were used in our study. The images were opened at the personal workstation Horos Medical Image Viewer (Version 3.0, USA) program and processed with 3D Curved Multiplanar Reconstruction (MPR). The line passing through the nasion and inion points of the images obtained as a result of the process was determined, and all images were brought to the orthogonal plane. Later, the images were overlapped and saved in JPEG format with 100% magnification. JPEG images saved were converted into TPS format, and 21 homologous landmarks were placed. Generalized Procrustes Analysis (GPA) and Principal Component Analysis (PCA) were applied to the coordinates of landmarks, and shape variations and dimensionality were corrected by gathering the images to the center of gravity. Next, Linear Discriminant Analysis (LDA) was applied to the coordinates, the dimensionality of which was corrected. Results: The study found that 74.465% of the coordinates of 21 homologous landmarks gathered to the center of gravity could be explained with the first three PCs. As a result of the LDA applied to these coordinates, a gender prediction rate of 86.5% was obtained. In addition, a slight difference was found between the GPA sum of squares and the tangent sum of squares (0.57). Conclusion: The images of the cranium obtained from CT showed a high dimorphism by geometric morphometry analysis.

Quaternion attention-based JND model for macrophotography image watermarking

With the advancement of imaging technology, macrophotography images (MPIs) have become a popular research topic. Unlike natural images, MPIs often feature sharp foregrounds and blurred backgrounds, leading to distinct perceptual characteristics in estimation. As the number of MPIs grows rapidly, concerns over image quality and security increase. Robust watermarking techniques have been introduced to address these challenges. Just Noticeable Difference (JND) has been widely used in quantization-based watermarking frameworks. However, existing JND models handle each image area with a single-level perceptual attention. Visual attention in Quaternion Discrete Wavelet Transform (QDWT), which can reflect the Multi-level perceptual attention feature. Therefore, we propose a new method called Quaternion Attention-based Just Noticeable Difference model for MPIs Watemarking (QAJnd-MW) for watermarking MPIs. This method uses visual attention mechanisms, recognizing that the HVS is more sensitive to attention regions. We generate a masking effect in the JND field. The input image undergoes QDWT to explore multi-scale features. The multi-scale feature maps, with multi-directional luminance and multi-channel color, help create local and global attention maps, which are fused to form the final attention map. Specifically, considering both attention-based masking effects, the quaternion attention-guided JND model is designed for a robust MPI watermarking framework, aiming to further improve MPI security. Extensive experiments on the MP2020 and Blur Detection datasets show that the proposed model significantly improves robustness against JPEG compression attacks, reducing the bit error rate (BER) by up to 12%. Additionally, the model performs well against other attacks, such as those in online social networks, with lower BER than current state-of-the-art techniques.

Reversible data hiding in JPEG images based on improved frequency selection and mapping strategy

JPEG images have become ubiquitous in our daily lives due to their ability to strike a balance between visual quality and file size. Consequently, reversible data hiding (RDH) schemes for JPEG images have emerged. However, existing methods by modifying discrete cosine transform (DCT) coefficients still have room for improvement in terms of distortion reduction and file size preservation. This paper proposes a novel two-dimensional (2D) histogram mapping strategy for RDH in JPEG images. Firstly, the quantized DCT coefficient blocks are sorted based on the count of zero alternating current (AC) coefficients in each block. This enables the estimation of distortion in non-zero AC coefficients at different locations within the block, facilitating the selection of frequency coefficients with minimal distortion for embedding. Additionally, leveraging statistical characteristics, six mapping types with different embedding efficiency are created and the improved 2D histogram mapping method is proposed. Furthermore, The unit distortion-increase ratio (UDIR) is employed as a comprehensive evaluation metric. Extensive experiments are conducted on four representative images and two widely-used image databases to validate the proposed scheme. The results consistently demonstrate superior visual quality, minimal file size increase (FSI), and higher UDIR in comparison to state-of-the-art RDH schemes for JPEG images.

Mastering Deepfake Detection: A Cutting-edge Approach to Distinguish GAN and Diffusion-model Images

Detecting and recognizing deepfakes is a pressing issue in the digital age. In this study, we first collected a dataset of pristine images and fake ones properly generated by nine different Generative Adversarial Network (GAN) architectures and four Diffusion Models (DM). The dataset contained a total of 83,000 images, with equal distribution between the real and deepfake data. Then, to address different deepfake detection and recognition tasks, we proposed a hierarchical multi-level approach. At the first level, we classified real images from AI-generated ones. At the second level, we distinguished between images generated by GANs and DMs. At the third level (composed of two additional sub-levels), we recognized the specific GAN and DM architectures used to generate the synthetic data. Experimental results demonstrated that our approach achieved more than 97% classification accuracy, outperforming existing state-of-the-art methods. The models obtained in the different levels turn out to be robust to various attacks such as JPEG compression (with different quality factor values) and resize (and others), demonstrating that the framework can be used and applied in real-world contexts (such as the analysis of multimedia data shared in the various social platforms) for support even in forensic investigations to counter the illicit use of these powerful and modern generative models. We are able to identify the specific GAN and DM architecture used to generate the image, which is critical in tracking down the source of the deepfake. Our hierarchical multi-level approach to deepfake detection and recognition shows promising results in identifying deepfakes allowing focus on underlying task by improving (about 2% on the average) standard multiclass flat detection systems. The proposed method has the potential to enhance the performance of deepfake detection systems, aid in the fight against the spread of fake images, and safeguard the authenticity of digital media.

Reversible Data Hiding in Shared JPEG Images

Reversible data hiding (RDH) in encrypted images has emerged as an effective technique for securely storing and managing confidential images in the cloud. However, most RDH methods in shared images (RDHSI) are designed for uncompressed images and cannot be applied for JPEG images. To address this issue, we propose a novel RDH in shared JPEG images. Our method consists of JPEG image sharing and data hiding in JPEG shares, which are both conducted on JPEG bit-stream. Specifically, the DC appended bits (DCA) and AC appended bits (ACA) derived from the original JPEG bit-stream are shared by ( \(k\) , \(n\) ) threshold Chinese remainder theorem based secret sharing (CRTSS) with two different constraints, one for DC sharing and another for AC sharing. The constraint of DC sharing ensures that the DC coefficient shares do not overflow. The constraint of AC sharing ensures the sizes of ACA shares are less than the sizes of the original ACA so that the embedding room can be vacated from each shared JPEG bit-stream. Each data-hider can embed the secret data into the personal JPEG share. The original JPEG image can be recovered losslessly from any \(k\) JPEG shares. The proposed sharing and data hiding are both well compatible with the JPEG standard. Experimental results demonstrate that the proposed method not only well preserves the file size whether the JPEG shares or marked JPEG shares, but also achieves outstanding security performance and a high embedding capacity.

Investigation of the shape of goat (capra hircus) astragalus via geometric morphometry method

The aim of this study is to determine the effect of sex on shape of goat astragalus via geometric morphometry method. A total of 37 astragalus bone samples collected from 16 female and 21 male goats were used as material. Bone samples were cleaned from skin and soft tissue and then macerated by boiling. Astragalus bone samples were photographed from a distance of 20 cm dorsally by focusing on the center of the bone. The photographs were transferred to the computer in JPEG format. Principal Component Analysis, Regression Analysis, Canonical Variate Analysis, and Discriminant Function Analysis were conducted using the Cartesian coordinate values, which were obtained by homologous landmark marking. The first two principal components accounted for 75.967% of the total shape variation. Shape variation was determined in different regions of the astragalus. According to the scatter plot of male and female individuals, male individuals were completely placed within the confidence interval ellipse of female individuals. It was found that allometric effect on the shape of astragalus bone was not statistically significant. As a result of Canonical Variate Analysis, mahalanobis and procrustes distances were detected as 2.9216 (p

RIHINNet: A robust image hiding method against JPEG compression based on invertible neural network

RIHINNet: A robust image hiding method against JPEG compression based on invertible neural network

Generative adversarial networks with deep blind degradation powered terahertz ptychography

Ptychography is an imaging technique that uses the redundancy of information generated by the overlapping of adjacent light regions to calculate the relative phase of adjacent regions and reconstruct the image. In the terahertz domain, in order to make the ptychography technology better serve engineering applications, we propose a set of deep learning terahertz ptychography system that is easier to realize in engineering and plays an outstanding role. To address this issue, we propose to use a powerful deep blind degradation model which uses isotropic and anisotropic Gaussian kernels for random blurring, chooses the downsampling modes from nearest interpolation, bilinear interpolation, bicubic interpolation and down-up-sampling method, and introduces Gaussian noise, JPEG compression noise, and processed detector noise. Additionally, a random shuffle strategy is used to further expand the degradation space of the image. Using paired low/high resolution images generated by the deep blind degradation model, we trained a multi-layer residual network with residual scaling parameters and dense connection structure to achieve the neural network super-resolution of terahertz ptychography for the first time. We use two representative neural networks, SwinIR and RealESRGAN, to compare with our model. Experimental result shows that the proposed method achieved better accuracy and visual improvement than other terahertz ptychographic image super-resolution algorithms. Further quantitative calculation proved that our method has significant advantages in terahertz ptychographic image super-resolution, achieving a resolution of 33.09 dB on the peak signal-to-noise ratio (PSNR) index and 3.05 on the naturalness image quality estimator (NIQE) index. This efficient and engineered approach fills the gap in the improvement of terahertz ptychography by using neural networks.