Source Camera Research Articles

Identifying the Digital Camera from Natural Images Using Residual Noise and the Jensen–Shannon Divergence

Regarding the problem of digital camera identification, many methods have been proposed, and for several of them, their effectiveness has been verified on the basis of disputed flat images. However, in real cases the disputed images are natural images, rather than flat images. In that case, several of the already proposed methods are not effective. Hence, in this paper, a method is proposed for the digital camera identification from natural images based on the statistical comparison between the residual noise in the natural disputed images and the fingerprint defined for the eligible digital cameras. In the reported case studies, the HDR database provided by the Communications and Signal Processing Laboratory of University of Florence is used to select a set of eligible digital cameras, and from this image database, for each digital camera, a set of disputed flat images, a set of disputed natural images, and a set of flat reference images were selected. Thus, the fingerprint of each digital camera was calculated from the probability density function (PDF) of the photo-response nonuniformity (PRNU) extracted from its reference images. Therefore, in order to identify the source digital camera of a natural disputed image, the Jensen–Shannon divergence (JSD) was implemented to statistically compare the PRNU-based fingerprint of each eligible source camera against the noise residual of that disputed image. The proposed method has a similar effectiveness to methods based on the peak-to-correlation energy or the Kullback–Leibler divergence when the disputed images are flat images and the PRNU is considered, but it is significantly more effective than those methods when the disputed images are natural images.

Performance enhancement of PRNU-based source identification for smart video surveillance

This paper introduces a current signal-based source verification (SSV) system for images in video surveillance networks including the cloud. Using a signal, the well-known photo-response non-uniformity (PRNU) can be used, which is unique and intrinsic in every digital image taken by a source camera, like fingerprints. The SSV system using PRNU has proved before to be useful for reliable video source identification in both network- and cloud-based video surveillance. However, in the era of smart living, security video systems have become part of the IoT devices which typically have limited resources such as low computation, power, storage and memory. To address these problems in the IoT applications, the effects of I-frames only and infra-red night scenes are studied as well as two proposed approaches for the SSV system. Then a hybrid version of the SSV scheme is further suggested, in combination with the best approach using averaged noise residues (for reduced false positive rate), and a recent technique using spatial domain averaged frames (for reduced computational complexity). The enhanced performance of the improved SSV system for smart video surveillance has been verified through tests.

Multi-DS Strategy for Source Camera Identification in Few-Shot Sample Data Sets

Source camera identification (SCI) is an intriguing problem in digital forensics, which identifies the source device of given images. However, most existing works require sufficient training samples to ensure performance. In this work, we propose a method based on semi-supervised ensemble learning (multi-DS) strategy, which extends labeled data set by multi-distance-based clustering strategy and then calibrate the pseudo-labels through a self-correction mechanism. Next, we iteratively perform the calibration-appending-training process to improve our model. We design comprehensive experiments, and our model achieves satisfactory performance on benchmark public databases (i.e., Dresden, VISION, and SOCRatES).

A PNU-Based Methodology to Improve the Reliability of Biometric Systems.

Face recognition is an important application of pattern recognition and image analysis in biometric security systems. The COVID-19 outbreak has introduced several issues that can negatively affect the reliability of the facial recognition systems currently available: on the one hand, wearing a face mask/covering has led to growth in failure cases, while on the other, the restrictions on direct contact between people can prevent any biometric data being acquired in controlled environments. To effectively address these issues, we designed a hybrid methodology that improves the reliability of facial recognition systems. A well-known Source Camera Identification (SCI) technique, based on Pixel Non-Uniformity (PNU), was applied to analyze the integrity of the input video stream as well as to detect any tampered/fake frames. To examine the behavior of this methodology in real-life use cases, we implemented a prototype that showed two novel properties compared to the current state-of-the-art of biometric systems: (a) high accuracy even when subjects are wearing a face mask; (b) whenever the input video is produced by deep fake techniques (replacing the face of the main subject) the system can recognize that it has been altered providing more than one alert message. This methodology proved not only to be simultaneously more robust to mask induced occlusions but also even more reliable in preventing forgery attacks on the input video stream.

Tampering detection and localization base on sample guidance and individual camera device convolutional neural network features

AbstractThis paper proposes an image tampering detection algorithm based on sample guidance and individual camera device's convolutional neural network (CNN) features (SGICD‐CF) to address the challenges in the authenticity and integrity of images. Due to the development of the digital image processing technology, which makes image editing and processing, image tampering and forgery easy and lot simplified, thus solving the problem of image tamper detection, to maintain information security. The principle of SGICD‐CF assumes that pixels of the pristine image come from a single camera device, but on the contrary, if an image to be tested is spliced by multiple images from different cameras, then the pixels from the multiple camera devices will be detected. SGICD‐CF divides the image to be tested into 64 × 64 pixel image patches, extracts the camera‐related features and some camera model‐related information of image patches by source camera identification network (SCI‐Net) which is proposed by us, and obtains the classification confidence degree of the image patch. Furthermore, it determines whether the image patch contains foreign pixels according to the obtained confidence degree and finally determines whether the image was tampered according to the classification results of all the image patches, thus locating the tampered area. However, the experimental results show that SGICD‐CF can detect and locate the tampered area of an image accurately and our methods have a better performance than other existing methods. Our algorithm can achieve an average correct rate of 0.855 on the synthetic data set based on Dresden, which is higher than other existing detection methods.

What Can We Learn from Depth Camera Sensor Noise?

Although camera and sensor noise are often disregarded, assumed negligible or dealt with in the context of denoising, in this paper we show that significant information can actually be deduced from camera noise about the captured scene and the objects within it. Specifically, we deal with depth cameras and their noise patterns. We show that from sensor noise alone, the object’s depth and location in the scene can be deduced. Sensor noise can indicate the source camera type, and within a camera type the specific device used to acquire the images. Furthermore, we show that noise distribution on surfaces provides information about the light direction within the scene as well as allows to distinguish between real and masked faces. Finally, we show that the size of depth shadows (missing depth data) is a function of the object’s distance from the background, its distance from the camera and the object’s size. Hence, can be used to authenticate objects location in the scene. This paper provides tools and insights into what can be learned from depth camera sensor noise.

Online Event-Based Insights into Unsteady Flows with TrackAER

We present a novel event-based quantitative flow visualization system, TrackAER, capable of continuously reconstructing, rendering and recording particle tracks in large test volumes and without limitations to the measurement duration. Multiple event-based cameras are synchronized and calibrated to produce independent and asynchronous, yet temporally co-registered data streams of flow tracers. Subsequently, these data streams are merged into time-resolved 3D particle tracks using photogrammetric techniques. Due to the operating principle of event cameras, the flow scenery is reduced to moving objects only, which effectively compresses the data stream at the camera source. In combination with an efficient data processing pipeline, the measurement system operates online. That is, the investigated flow field is reconstructed and rendered in an interactive 3D viewer without noticeable time lag. The data processing approach follows a “per-event” paradigm and is achieved in real-time, enabling the immediate observation and analysis of both, transient and long duration flow features. Specific issues resulting from event-based, frame-free processing are discussed as well as the advantages and limitations of event cameras. Exemplary results of two different test cases are provided to demonstrate the utility of the TrackAER system in- and outside a wind tunnel, where Lagrangian particle track information is displayed in a virtual scene together with extracted quantitative information such as local flow velocities.

A novel image dataset for source camera identification and image based recognition systems

Multimodal emotion recognition has attracted a great deal of attention in recent years, with new interesting applications now being considered. One promising application is in the digital image forensics fields where, for example, it gives the possibility to automatically highlight subjects that are in pain, in digital images under examination, by analyzing their facial expressions. However, finding an image that represents a possible crime leaves the problem of identifying the device used to take the image open. Such a problem has been addressed by Source Camera Identification algorithms (SCI, for short). These algorithms analyze some features hidden in a target image to find traces left by the sensor that captured the image. A particularly challenging case is when the candidate source cameras for an image under investigation are of the same manufacturer and model. A fair and universal assessment of these algorithms is only possible if standard datasets are used for their benchmarking. However, our comprehensive analysis has shown that the majority of the datasets proposed so far contain a collection of images taken with different types of cameras, mostly smartphones. We fill this gap by presenting UNISA2020, a novel image dataset that contains a large collection of real-world images taken with multiple conventional digital cameras of the same type. The images in our dataset have been assembled so as to avoid artifacts that could negatively affect the identification process. To validate our dataset, we also performed a comparative experimental analysis to investigate the performance of an SCI reference algorithm when running on our dataset as well as on other SCI standard datasets.

Camera model identification based on forensic traces extracted from homogeneous patches

A crucial challenge in digital image forensics is to identify the source camera model used to generate given images. This is of prime importance, especially for Law Enforcement Agencies in their investigations of Child Sexual Abuse Material found in darknets or seized storage devices.In this work, we address this challenge by proposing a solution that is characterized by two main contributions. It relies on the extraction of rather small homogeneous regions that we extract very efficiently from the integral image, and on a hierarchical classification approach with convolutional neural networks as the underlying models. We rely on homogeneous regions as they contain camera traces that are less distorted than regions with high-level scene content. The hierarchical approach that we propose is important for scaling up and making minimal modifications when new cameras are added. Furthermore, this scheme performs better than the traditional single classifier approach. By means of thorough experimentation on the publicly available Dresden data set, we achieve an accuracy of 99.01% with 5-fold cross-validation on the ‘natural’ subset of this data set. To the best of our knowledge, this is the best result ever reported for Dresden data set.

Source Camera Device Identification from Videos

Source camera identification is an important and challenging problem in digital image forensics. The clues of the device used to capture the digital media are very useful for Law Enforcement Agencies (LEAs), especially to help them collect more intelligence in digital forensics. In our work, we focus on identifying the source camera device based on digital videos using deep learning methods. In particular, we evaluate deep learning models with increasing levels of complexity for source camera identification and show that with such sophistication the scene-suppression techniques do not aid in model performance. In addition, we mention several common machine learning strategies that are counter-productive in achieving a high accuracy for camera identification. We conduct systematic experiments using 28 devices from the VISION data set and evaluate the model performance on various video scenarios—flat (i.e., homogeneous), indoor, and outdoor and evaluate the impact on classification accuracy when the videos are shared via social media platforms such as YouTube and WhatsApp. Unlike traditional PRNU-noise (Photo Response Non-Uniform)-based methods which require flat frames to estimate camera reference pattern noise, the proposed method has no such constraint and we achieve an accuracy of 72.75 pm 1.1 % on the benchmark VISION data set. Furthermore, we also achieve state-of-the-art accuracy of 71.75% on the QUFVD data set in identifying 20 camera devices. These two results are the best ever reported on the VISION and QUFVD data sets. Finally, we demonstrate the runtime efficiency of the proposed approach and its advantages to LEAs.

Joint generative and camera-aware clustering for unsupervised domain adaptation on person re-identification

Existing supervised person re-identification (Re-ID) methods demonstrate excellent performance. However, their performances suffer degradation when tested on an unseen different distributed domain. Generative adversarial network-based (GAN-based) and clustering- or pseudolabel-based methods are proposed to alleviate this problem. Due to the transfer scheme and the ignorance of correlations between the style-transferred and the original target images, the performance of GAN-based methods is unsatisfactory. We resolve these problems by jointly employing a generative strategy and performing camera-aware clustering for the target domain. Style-transferred images are generated from source cameras to target cameras, and then they are merged into the target domain selectively after exploiting their domain-specific discriminative information. To reduce the noise in generated images, we propose a domain-level boundary separation loss to group the transferred images and push them away from the original target images. The camera-level neighborhood-based clustering is proposed to learn well-clustered features in a camera-aware manner. Extensive experiments on two commonly used person Re-ID datasets demonstrate that our proposed method can achieve state-of-the-art performance.

Comparative Analysis of Optoelectronic Accuracy in the Laboratory Setting Versus Clinical Operative Environment: A Systematic Review

Study Design:Systematic review.Objectives:The optoelectronic camera source and data interpolation process serve as the foundation for navigational integrity in robotic-assisted surgical platforms. The current systematic review serves to provide a basis for the numerical disparity observed when comparing the intrinsic accuracy of optoelectronic cameras versus accuracy in the laboratory setting and clinical operative environments.Methods:Review of the PubMed and Cochrane Library research databases was performed. The exhaustive literature compilation obtained was then vetted to reduce redundancies and categorized into topics of intrinsic accuracy, registration accuracy, musculoskeletal kinematic platforms, and clinical operative platforms.Results:A total of 465 references were vetted and 137 comprise the basis for the current analysis. Regardless of application, the common denominators affecting overall optoelectronic accuracy are intrinsic accuracy, registration accuracy, and application accuracy. Intrinsic accuracy equaled or was less than 0.1 mm translation and 0.1 degrees rotation per fiducial. Controlled laboratory platforms reported 0.1 to 0.5 mm translation and 0.1 to 1.0 degrees rotation per array. Accuracy in robotic-assisted spinal surgery reported 1.5 to 6.0 mm translation and 1.5 to 5.0 degrees rotation when comparing planned to final implant position.Conclusions:Navigational integrity and maintenance of fidelity of optoelectronic data is the cornerstone of robotic-assisted spinal surgery. Transitioning from controlled laboratory to clinical operative environments requires an increased number of steps in the optoelectronic kinematic chain and error potential. Diligence in planning, fiducial positioning, system registration and intra-operative workflow have the potential to improve accuracy and decrease disparity between planned and final implant position.

Building Camera Source Identification Image Database through Croudsourcing

디지털카메라 사용의 편의성과 활용이 증가하면서 촬영된 이미지로부터 촬영에 사용된 소스 카메라를 식별하는 다양한 연구가 진행되고 있다. 하지만 이 분야에도 딥러닝 기술을 본격적으로 활용하기 시작하면서 제안하는 알고리즘을 개발하거나 검증하기 위한 실험에 필요한 이미지 데이터베이스가 부족한 실정이다. 또한, 기존 이미지 데이터베이스는 최신 카메라 모델을 지원하지 않거나, 디지털카메라와 스마트폰 카메라 중 하나에 대한 데이터를 제공하는 데 그쳐 현실성이 부족하다. 본 논문에서는 크라우드소싱 방식을 통해 수집된 Sangmyung Image Database (SMDB)를 대한 유효성 검증 결과를 보여주고 데이터를 공유하고자 한다. SMDB는 최신 모델의 9개의 디지털카메라와 26개의 스마트폰 카메라로 촬영된 약 12,000장의 이미지로 구성되었으며, Flatfield, Dark, Natural 이미지로 구분되어, 이미지 센서의 노이즈 패턴을 분석하거나 소스 카메라를 식별하는 데 사용할 수 있다. 또한 SMDB는 크라우드소싱을 통해 불특정 다수의 일반인에 의해 수집되었음에도 FODB, VISION, Dresden 등 기존 이미지 데이터베이스와 유사한 수준의 카메라 소스 식별 정확도를 보여주어 수집된 데이터에 대한 절차의 신뢰성과 결과의 정확성 모두를 만족시켜준다.

Virtual sample generation for few-shot source camera identification

The Source Camera Identification (SCI) has achieved remarkable success. However, existing approaches require sufficiently large training sets for high performance on accuracy and robustness. For maintaining high performance given small training sets, we propose a semi-supervised, Mega-Trent-Diffusion (MTD) method to generate virtual samples, such that the training sets can be expanded and unlabeled samples can be fully utilized as well. The stability of our method is improved using ensemble learning. Our theoretical analysis and experiments corroborate the effectiveness of our method beyond others when few-shot is given.

A Deep Learning Framework for Image Authentication: An Automatic Source Camera Identification Deep-Net

A Deep Learning Framework for Image Authentication: An Automatic Source Camera Identification Deep-Net

Digital Twins by Physical Education Teaching Practice in Visual Sensing Training System

The paper expects to improve the efficiency and intelligence of somatosensory recognition technology in the application of physical education teaching practice. Firstly, the combination of induction recognition technology and the Internet is used. Secondly, through the Kinect sensor, bone data are acquired. Finally, the hidden Markov model (HMM) is used to simulate the experimental data. On the simulation results, a gait recognition algorithm is proposed. The gait recognition algorithm is used to identify the motion behaviour, and the results are displayed in the Web (World Wide Web) end built by the cloud server. Meantime, in view of the existing problems in the practice of physical education, combined with the establishment and operation of the Digital Twins (DTs) system, the camera source recognition architecture is carried out since the twin network and the two network branches share weights. This paper analyses these problems since the application of somatosensory recognition technology and puts forward the improvement methods. For the single problem of equipment in physical education, this paper puts forward the monitoring and identification function of the cloud server. It is to transmit data through Hypertext Transfer Protocol (HTTP) and locate and collect data through a monitoring terminal. For the lack of comprehensiveness and balance of sports plans, this paper proposes a scientific training plan and process customization based on Body Mass Index (BMI), analyses real-time data in the cloud, and makes scientific customization plans according to different students’ physical conditions. Moreover, 25 participants are invited to carry out the exercise detection and analysis experiment, and the joint monitoring of their daily movements is tested. This process has completed the design of a feasible and accurate platform for information collection and processing, which is convenient for managers and educators to comprehensively and scientifically master and manage the physical level and training of college students. The proposed method improves the recognition rate of the camera source to some extent and has important exploration significance in the field of action recognition.

A New Forensic Video Database for Source Smartphone Identification: Description and Analysis

In recent years, the field of digital imaging has made significant progress, so that today every smartphone has a built-in video camera that allows you to record high-quality video for free and without restrictions. On the other hand, rapidly growing internet technology has contributed significantly to the widespread use of digital video via web-based multimedia systems and mobile smartphone applications such as YouTube, Facebook, Twitter, WhatsApp, etc. However, as the recording and distribution of digital videos have become affordable nowadays, security issues have become threatening and spread worldwide. One of the security issues is identifying source cameras on videos. There are some new challenges that should be addressed in this area. One of the new challenges is individual source camera identification (ISCI), which focuses on identifying each device regardless of its model. The first step towards solving the problems is a popular video database recorded by modern smartphone devices, which can also be used for deep learning methods that are growing rapidly in the field of source camera identification. In this paper, a smartphone video database named Qatar University Forensic Video Database (QUFVD) is introduced. The QUFVD includes 6000 videos from 20 modern smartphone representing five brands, each brand has two models, and each model has two identical smartphone devices. This database is suitable for evaluating different techniques such as deep learning methods for video source smartphone identification and verification. To evaluate the QUFVD, a series of experiments to identify source cameras using a deep learning technique are conducted. The results show that improvements are essential for the ISCI scenario on video.

Open-Set Source Camera Device Identification of Digital Images Using Deep Learning

Source camera identification plays an important role in forensics investigations on images. It is a forensic problem of linking an image in question to the camera used to capture it. Several source identification techniques have been developed in the literature since this may be a facilitating tool that help to trace back the images to the camera device held by the accused in various forensic applications. However, one of the key disadvantages is that the existing techniques fail if the image in question was taken by a new camera that is not used in the training process. Under a real-world forensic scenario, it is not possible to presume that each image being analyzed comes from one of the cameras used to train the source identification system. To address this issue, we propose a data-driven system based on convolutional neural network to identify the source camera device in an open-set scenario. The experimental results on various sets of cameras show that it is possible to leverage the data-driven model as the feature extractor paired with an open-set classifier to trace back the images to the open-set cameras. The results show that the proposed system outperforms the state-of-the-art techniques in identifying the exact device that are never seen before with considerably high accuracy and is resilient to unknown post-processing applied by the social network platforms. Moreover, the experimental results demonstrate the good generalization capability of the proposed system in extracting the source information, making it more suitable for open-set scenarios.

Open-Set source camera identification based on envelope of data clustering optimization (EDCO)

The task of source camera identification is devoted to linking an image to the source camera model and plays a significant role in forensics. Nevertheless, with the ongoing development of new camera models, it is difficult to keep a model database up to date, giving rise to the open-set problem. To deal with this problem, we propose a novel approach based on the envelope of data clustering optimization (EDCO). The new EDCO scheme can identify the camera model regardless of whether or not it is included in the database. The experimental results prove that EDCO efficiently separates unknown source images from known source images and links the query image identified as known with the source camera model. When the dataset is expanded with the new camera model, EDCO only needs to train the new model instead of retraining with all models together, which greatly improves the scalability. Compared with the state of the art, our method can effectively distinguish between images from known and unknown camera models, even in extreme cases.

PNU Spoofing: a menace for biometrics authentication systems?

• The analysis of the PNU noise is the most popular approach for identifying the camera used to take an image under scrutiny. • PNU is also used in other domains, like iris-based authentication and health monitoring based on bio-signal processing. • Several spoofing techniques exist for deceiving identification algorithms PNU-based and undermining the integrity of images. • We show that spoofed images retain traces of their originating device, these allow to correctly perform the identification. The Pixel Non-Uniformity noise (PNU noise, for short) is a characteristic noise of digital camera sensors that has been originally used a mean to perform Source Camera Identification (SCI), that is, to identify the digital camera that has been used to take an image under scrutiny. Actually, its usage has been extended to other relevant application domains, such as carrying out sensor identification in iris biometrics , to resolve the integrity of a biometric authentication system , in health monitoring systems based on bio-signal processing, and to prevent health fraud scams. As a consequence of this popularity, several counter-forensics techniques have been proposed in the scientific literature to deceive this identification approach and, thus, putting at risk its effectiveness. The goal of this paper is to experiment with some of these counter-forensics techniques and to assess whether they are able or not to deceive the PNU-based identification process. We focus our attention on the particular case where a target image is modified so as to appear as been taken by a different device than the original one (i.e., spoofing). The results of our experiments show that this kind of spoofing is apparently successful in deceiving the PNU-based identification process. However, we also show that the spoofed images retain traces of their originating camera. By leveraging on this information, it is still possible to trace back the device used to take that image. This finding may have a deep impact on the usage of Pixel Non Uniformity (PNU) noise for identification tasks, such as the ones carried out by biometric recognition and monitoring systems, and on the possibility to really cheat them by means of counter-forensics techniques.