Zoran Duric | George Mason University (original) (raw)

Papers by Zoran Duric

This chapter concerns problems of learning patterns in images and image sequences, and using the ... more This chapter concerns problems of learning patterns in images and image sequences, and using the obtained patterns for interpreting new images. The chapter concentrates on three problem areas: (i) semantic interpretation of color images of outdoor scenes, (ii) detection of blasting caps in x-ray images of luggage, and (iii) recognizing actions in video image sequences. It discusses the image formation processes in these problem areas, and the choices of representation spaces used in our approaches to solving these problems. The results presented indicate the advantages of applying machine learning to vision.

... i FOC i ---~ , TI V NAOR -£2 V ... The two singular points (the AOR and the NAOR) at which th... more ... i FOC i ---~ , TI V NAOR -£2 V ... The two singular points (the AOR and the NAOR) at which the optical flow is zero correspond to the points where ~ and -fi pierce the image egosphere. as a singular point of type L If cos 0 = 0 (0 = zr/2) we have the angle Pc ~00 = arccos -- (10) p ...

Advances in information security, 2001

Welcome to the first volume of ADVANCES IN INFORMATION SECURITY. The goals of this series are to ... more Welcome to the first volume of ADVANCES IN INFORMATION SECURITY. The goals of this series are to establish the state of the art, set the course for future research in information security, and to serve as a central source of reference for information security research and developments. The scope of this series includes not only all aspects of computer and network security, but related areas such as fault tolerance and software assurance. The series aims to publish thorough and cohesive overviews on specific topics in Information Security, as well as works that are larger in scope than survey articles and that will contain more detailed background information. The series also provides a single point of coverage of advanced and timely topics and a forum for topics that may not have reached a level of maturity to warrant a comprehensive textbook. SUSHlL JAJODIA Consulting Editor To my parents Bill and Carolyn, wife Ann-Marie, and son William.-NFJ To my wife Sladjana, and my children Petar and Sonja.-ZD To my parents.-SJ Contents LIST OF FIGUR.ES .

We describe a method of recognizing hand gestures from hand silhouettes. Given the silhouette of ... more We describe a method of recognizing hand gestures from hand silhouettes. Given the silhouette of a hand, we compute its convex hull and extract the deficits of convexity corresponding to the differences between the hull and the silhouette. The deficits of convexity are normalized by rotating them around the edges shared with the hull. To learn a gesture, the deficits from a number of examples are extracted and normalized. The deficits are grouped by similarity which is measured by the relative overlap using k-means clustering. Each cluster is assigned a symbol and represented by a template. Gestures are represented by string of symbols corresponding to the nearest neighbors of the deficits. Distinct sequences of symbols corresponding to a given gesture are stored in a dictionary. Given an unknown gesture, its deficits of convexity are extracted and assigned the corresponding sequence of symbols. This sequence is compared with the dictionary of known gestures and assigned to the class to which the best matching string belongs. We used our method to design a gesture interface to control a web browser. We tested our method on five different subjects and achieved a recognition rate of 92%-99%.

3D models of buildings are used in many applications such as location recognition, augmented real... more 3D models of buildings are used in many applications such as location recognition, augmented reality, virtual training and entertainment. Creating models of buildings automatically is a longstanding goal in computer vision research. Many current applications rely on manual creation of models using images and a 3D authoring tool. While more automated approaches exist, they typically are inefficient, require dense imagery, other sensor data, or frequent manual interventions. The focus of this thesis is to automate and increase the efficiency of 3D model creation from image collections. Matching sets of images to each other is a frequent step in 3D model building. In many applications image matching must be done hundreds or thousands of times. Thus, any increase in matching efficiency will be multiplied hundreds or thousands of times when used in these applications. This dissertation presents a new image matching method that achieves greater efficiency by using the fact that images taken from similar viewing angles are approximately related by an affine transformation. An affine transformation models translation, rotation and non-isotropic scaling between image pairs. When images are related by an affine transformation ratios of areas of corresponding shapes are invariant. The method uses this invariant to fit an affine transformation model to a set of putative matches and detect incorrect matches. Methods assuming global and local affine transformation models were created. The first assumes a single global affine transformation between each image pairs. The second method imposes a structure on the feature points to cluster features in a local region. The method then fits different affine models to each cluster. Both methods were evaluated using sets of synthetic matches with varying percentages of incorrect matches, localization error and rotation. Additionally, the methods were applied to a large publicly available image database and the results were compared to several recent model fitting methods. The results show the best affine method using local regions maintains equivalent accuracy and is consistently more efficient than current state of the art methods. When creating and using 3D models, it is often important to predict if images taken from specific locations will match existing images in the model. Image matching prediction is used to evaluate image sets for vision-based location recognition and augmented reality applications. This dissertation presents a new way to predict if images will match by measuring affine distortion. Distortion is measured by projecting features into a second image and computing the affine transformation between the corresponding feature regions. Feature distortion is computed from the skew, stretch and shear of the transformed region. Using the distortion measure for all features in an image pair, a distortion vector is created describing the image pair. Using the distortion vectors and the actual number of matches, a classifier is trained to predict the confidence that images will match. Results are presented that compare this method to other published approaches. The results demonstrate the affine distortion-based classifier predicts matching confidence more accurately than other published techniques. The classifier is also used to create a spatial model of locations around a building. The spatial model shows the confidence that a new image taken from a specific location and pose will match an existing set of images. Using this model, location recognition applications can determine how well they will work throughout the scene. The approach presented uses the classifier described above and more realistic location sampling to create a spatial map that is more accurate than other published approaches. Additionally, as part of this goal, the minimum set of images needed to cover the space around the building is computed. The approach uses structure from motion to create 3D information about the scene. Synthetic cameras are then created using approximate locations and directions from which people commonly take pictures. The affine distortion-based classifier is applied to compute the confidence that images from the synthetic cameras will match the existing set of images. Results are presented on a spatial map showing the confidence that new images captured at specific locations and poses will match the existing image set. Additionally, the minimal set of images needed to maintain the matching coverage is computed using a greedy set cover algorithm. The minimal set can be used to increase efficiency in applications that need to match new images to an existing set of images (e.g. location recognition, augmented reality and 3D modeling applications). Finally, a process is presented to validate the 3D information computed using structure from motion. Validation ensures that the data is precise and accurate enough to provide a realistic 3D model of the scene structure. Results…

Page 1. Information Hiding: Steganography and Watermarking Dr. Mohammed Al-Mualla and Prof. Hussa... more Page 1. Information Hiding: Steganography and Watermarking Dr. Mohammed Al-Mualla and Prof. Hussain Al-Ahmad Multimedia Communication and Signal Processing (MCSP) Research Group Etisalat College of Engineering POBox: 980, Sharjah, UAE ...

Springer eBooks, 2000

Many techniques for watermarking of digital images have appeared in numerous publications. Most o... more Many techniques for watermarking of digital images have appeared in numerous publications. Most of these techniques are sensitive to cropping and or a ne distortions e.g., rotation and scaling. In this paper we describe a method for the recovery of original size and appearance of images based on the concept of identi cation marks ngerprints"; the method does not require the use of the original" image, but only a small numberofsalient image points. We s h o w that, using our method, it is possible to recover original appearances of distorted images. The restored image can be used to recover embeddedwatermarks.

Frontiers in Rehabilitation Sciences

The analysis of functional upper extremity (UE) movement kinematics has implications across domai... more The analysis of functional upper extremity (UE) movement kinematics has implications across domains such as rehabilitation and evaluating job-related skills. Using movement kinematics to quantify movement quality and skill is a promising area of research but is currently not being used widely due to issues associated with cost and the need for further methodological validation. Recent developments by computationally-oriented research communities have resulted in potentially useful methods for evaluating UE function that may make kinematic analyses easier to perform, generally more accessible, and provide more objective information about movement quality, the importance of which has been highlighted during the COVID-19 pandemic. This narrative review provides an interdisciplinary perspective on the current state of computer-assisted methods for analyzing UE kinematics with a specific focus on how to make kinematic analyses more accessible to domain experts. We find that a variety of ...

2017 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), 2017

Activities of Daily Living (ADLs) are personal functional activities performed by individuals to ... more Activities of Daily Living (ADLs) are personal functional activities performed by individuals to carry out their daily lives, allowing them to live independently. Finding relationships between surface electromyograms (sEMG) measured in the arm and movements of the hand and wrist needed to perform ADLs can help address performance deficits and be exploited in designing myoelectric control systems. This paper reports on applying machine learning techniques to discover the electromyogram patterns present when using the hand to perform 47 typical fine motor functional activities used to accomplish ADLs. A Hidden Markov Model (HMM) combined with Random Forest (RF) classification is employed to learn the patterns needed to identify 10 second segments of continuous movement. The HMM/RF model was applied using two feature sets: one consisting entirely of sEMG signals, the other adding accelerometer data. Results show an accuracy improvement of the HMM/RF model over RF-only classification, and further improvement using the combined sEMG+ACC features, with a range of 74.99% to 84.09% and average of 79.06% for five subjects.

Interactions between humans and computers can be augmented via recognition of human affective and... more Interactions between humans and computers can be augmented via recognition of human affective and cognitive states by the machine participant. Signals emitted by the human face have long been of interest to researchers in a wide range of fields for their usefulness as state indicators. Biometrics related to the eye region are of particular interest in a range of applications—from identification to adaptive interfaces. This thesis explores the potential of static and dynamic eye region biometrics as unique indicators of fatigue and cognitive engagement, and the impact of ambiguous eye region behaviors (e.g., partial blinks and asymmetric eyelid movements) on this process. Fatigue and engagement were selected as representative affective and cognitive states given their frequent entwined presence in commonly occurring HCI scenarios. To facilitate analysis, we integrate a novel collection of dynamic, minimally-intrusive computer vision techniques for effective interpretation of eye blin...

Security and Communication Networks, 2016

Today’s world’s societies are becoming more and more dependent on open networks such as the Inter... more Today’s world’s societies are becoming more and more dependent on open networks such as the Internet – where commercial activities, business transactions, and government services are realized. This has led to the fast development of new cyber threats and numerous information security issues which are exploited by cyber criminals. The inability to provide trusted secure services in contemporary computer network technologies has a tremendous socio-economic impact on global enterprises as well as individuals. Moreover, the frequently occurring international frauds impose the necessity to conduct the investigation of facts spanning across multiple international borders. Such examination is often subject to different jurisdictions and legal systems. A good illustration of the previously mentioned is the Internet, which has made it easier to perpetrate traditional crimes. It has acted as an alternate avenue for the criminals to conduct their activities, and launch attacks with relative anonymity. The increased complexity of the communications and the networking infrastructure is making investigation of the crimes difficult. Traces of illegal digital activities are often buried in large volumes of data, which are hard to inspect with the aim of detecting offenses and collecting evidence. Nowadays, the digital crime scene functions like any other network, with dedicated administrators functioning as the first responders. This poses new challenges for law enforcement policies and forces the computer societies to utilize digital forensics to combat the increasing number of cybercrimes. Forensic professionals must be fully prepared in order to be able to provide court admissible evidence. To make these goals achievable, forensic techniques should keep pace with new technologies. In this special issue, we are delighted to present a selection of 14 papers, which, in our opinion, will contribute to the enhancement of knowledge in cyber crime. The collection of high-quality research papers provides a view on the latest research advances and results in the field of digital forensics and to present the development of tools and techniques which assist the investigation process of potentially illegal cyber activity. In the first paper, Cyberterrorism targeting the general public through social media, Nicholas Ayres and Leandros A. Maglaras investigate whether a mimetic malware could be a viable method of attack against a population with respect to cyberterrorism. The presented research shows that although people are, in general, aware of cyberterrorism on their current level of fear of being a potential target of attack is relatively low. However, when presented with such a threat, their level of fear increased. The obtained results prove that a targeted mimetic virus can indeed have an effect on a population and is a potential attack method for cyberterrorism. The paper emphasizes also the importance of social media as a vessel of propagation of such threat. Next, in the paper entitled Effectiveness of File-Based Deduplication in Digital Forensics Sebastian Neuner, Martin Schmiedecker, and Edgar Weippl focus on introducing improvements to the standardized forensic process to reduce the amount of storage requirement for forensic investigations by using file whitelisting and cross-device deduplication. Authors approach is shown to be particularly useful in cases where investigation relies on referenced files in the file system. In the exemplary use case authors prove that file deduplication and file whitelisting can be successfully utilized to achieve 78% size reduction compared to the full data set which means saving about 700 gigabytes of storage capacity. Jawwad Shamsi, Sherali Zeadally, Fareha Sheikh, and Angelyn Flowers in Attribution in Cyberspace: Techniques and Legal Implications argue that only a few known cybercrimes have been successfully attributed to the actual attacker. To improve this situation authors propose threelevel attribution framework to indicate various attributes and guidelines through which attribution can be instigated. The proposed framework is an initial step and in order to be successful it requires strong cooperation between different stake holders, government sponsored active cyber unit, existence of cyber laws, and cooperation among international community members. Next, two papers are focused on anomaly detection. In Evolutionary-based Packets Classification for Anomaly Detection in Web Layer, Rafał Kozik, Michał Choraś, and Witold Hołubowicz propose a novel detection method for modern web applications. First, authors observe that the majority of the state of the art solutions make an assumption about the packets’ content, or how the data inside the payload is serialized. Then they propose an evolutionary-based approach to unsupervised and automated packets segmentation. On the top of their approach, authors apply several variants of machine-learned classifiers and statistics to prove that…

2020 42nd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC)

Haptic virtual environments have been used to assess cognitive and fine motor function. For tasks... more Haptic virtual environments have been used to assess cognitive and fine motor function. For tasks performed in physical environments, upper extremity movement is usually separated into reaching and object manipulation phases using fixed velocity thresholds. However, these thresholds can result in premature segmentation due to additional trajectory adjustments common in virtual environments. In this work, we address the issues of premature segmentation and the lack of a measure to characterize the spatial distribution of a trajectory while targeting an object. We propose a combined relative distance and velocity segmentation procedure and use principal component analysis (PCA) to capture the spatial distribution of the participant’s targeting phase. Synthetic data and 3D motion data from twenty healthy adults were used to evaluate the methods with positive results. We found that these methods quantify motor skill improvement of healthy participants performing repeated trials of a haptic virtual environment task.

There is growing interest in the kinematic analysis of human functional upper extremity movement ... more There is growing interest in the kinematic analysis of human functional upper extremity movement (FUEM) for applications such as health monitoring and rehabilitation. Deconstructing functional movements into activities, actions, and primitives is a necessary procedure for many of these kinematic analyses. Advances in machine learning have led to progress in human activity and action recognition. However, their utility for analyzing the FUEM primitives of reaching and targeting during reach-to-grasp and reach-to-point tasks remains limited. Domain experts use a variety of methods for segmenting the reaching and targeting motion primitives, such as kinematic thresholds, with no consensus on what methods are best to use. Additionally, current studies are small enough that segmentation results can be manually inspected for correctness. As interest in FUEM kinematic analysis expands, such as in the clinic, the amount of data needing segmentation will likely exceed the capacity of existin...

The goal of steganography is to insert a message into a carrier signal so that it cannot be detec... more The goal of steganography is to insert a message into a carrier signal so that it cannot be detected by unintended recipients. Due to their widespread use and availability of bits that can be changed without perceptible damage of the original signal images, video, and audio are widespread carrier media. Steganalysis attempts to discover hidden signals in suspected carriers or at the least detect which media contain hidden signals. Therefore, an important consideration in steganography is how robust to detection is a particular technique. We review the existing steganography and steganalysis techniques and discuss their limitations and some possible research directions. Key words: Information hiding, steganography, steganalysis, watermarking 1

2021 43rd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC), 2021

Fine motor movement is a demonstrated biomarker for many health conditions that are especially di... more Fine motor movement is a demonstrated biomarker for many health conditions that are especially difficult to diagnose early and require sensitivity to change in order to monitor over time. This is particularly relevant for neurodegenerative diseases (NDs), including Parkinson's Disease (PD) and Alzheimer's Disease (AD), which are associated with early changes in handwriting and fine motor skills. Kinematic analysis of handwriting is an emerging method for assessing fine motor movement ability, with data typically collected by digitizing tablets; however, these are often expensive, unfamiliar to patients, and are limited in the scope of collectible data. In this paper, we present a vision-based system for the capture and analysis of handwriting kinematics using a commodity camera and RGB video. We achieve writing position estimation within 0.5 mm and speed and acceleration errors of less than 1.1%. We further demonstrate that this data collection process can be part of an ND screening system with a developed ensemble classifier achieving 74% classification accuracy of Parkinson's Disease patients with vision-based data. Overall, we demonstrate that this approach is an accurate, accessible, and informative alternative to digitizing tablets and with further validation has potential uses in early disease screening and long-term monitoring. Clinical relevance-This work establishes a more accessible alternative to digitizing tablets for extracting handwriting kinematic data through processing of RGB video data captured by commodity cameras, such as those in smartphones, with computer vision and machine learning. The collected data has potential for use in analysis to objectively and quantitatively differentiate between healthy individuals and patients with NDs, including AD and PD, as well as other diseases with biomarkers displayed in fine motor movement. The developed system has potential applications including providing widespread screening systems for NDs in low-income areas and resource-poor health systems, as well as an accessible form of disease long-term monitoring through telemedicine.

This paper describes a novel application of the MIST methodology to target detection in SAR image... more This paper describes a novel application of the MIST methodology to target detection in SAR images. Specifically, a polarimetric whitening filter and a constant false alarm rate detector are used to preprocess a SAR image; then the AQ15c learning program is applied to learn and detect targets. Encouraging and impressive experimental results are provided.

Vehicular safety depends on road geometry and surface conditions that provide tire-terrain tracti... more Vehicular safety depends on road geometry and surface conditions that provide tire-terrain traction. Roads are constructed with 3-D features consisting of super-elevation, grade, horizontal profile. A driving automaton that is aware of these parameters and surface friction can make better driving decisions at optimal speeds that maintain vehicular stability. Estimation of tire-terrain friction requires a continuous 3-D surface. In order to satisfy these requirements, we extend the Open Curved Regular Grid (OpenCRG) format with a C2 continuous surface that is dynamically generated at tire-terrain contact patches. We show that these estimates empower driving automatons to stably navigate curved elevated roads with super-elevated turns. We also show that sharing friction estimates using Basic Safety Messages (BSM) of Dedicated Short Range Communications (DSRC) can enable safety of following vehicles in control loss situations.

Marker-based imaging of human locomotion provides an extremely high level of accuracy, but it is ... more Marker-based imaging of human locomotion provides an extremely high level of accuracy, but it is quite intrusive and requires a significant amount of time for both the subject and the gait analyst. The purpose of automated gait analysis is to provide a means to analyze gait from video without the use of markers. Performing this analysis in an automated manner opens up a number of possibilities such as continuous analysis to monitor a course of treatment or to keep watch on the elderly population for changes in gait that might indicate a physical injury or change in mental condition. There are a number of factors that play into automated gait analysis. Different aspects (or determinants) of gait are active at different parts of the gait cycle. Therefore to provide analysis with respect to all determinants we must have a way of including gait cycle information. There is also the question of how the motion of the limbs can be analyzed. Limbs are constantly self-occluding, and issues such as poor contrast and loose clothing clutter the true motion of the limbs. Actual motion represents the ground truth of how the limb is moving, often times assumed in clinical analysis whereas in automated analysis this is not a given. Loose clothing may obscure actual limb motion, and motion analysis can only provide information about the apparent motion. For these reasons, we see automated gait analysis in some respects as complementary to marker based imaging of gait. It is not possible to have the same level of precision, but the availability and ease of this approach makes it much more applicable to a wider range of scenarios. In this thesis, we present an approach to automated gait analysis based on the motion of superpixels. We overlay the silhouette of the subject with a regular grid, where each grid cell represents a single superpixel. We overlay an additional superpixel to the top 13% of the body, which approximately corresponds to the head region. Human motion analysis is accomplished by analyzing the motion in each of the superpixels. We model the motion of the head using an affine motion model, which can account for a wide variety of valid motions that we can observe in the head (bending at the neck turning in a different direction, etc.). We use a three parameter "twist" motion model on the other regions of the body, which only models the translation and rotation in the superpixel. Finally, we build a representation of the data using independent components analysis (ICA). ICA provides a compact set of features describing the shape and motion of the body. We use independent components of motion to answer two different questions. The first: can we identify characteristics of the subject (i.e. gender and heel height) given shape and motion information. This is mostly important for identification in a soft biometrics sense. The second: can we identify a person that is walking in a similar manner using ICs. We demonstrate the robustness of the approach by taking it one step further by using ICs of each individual patch to compare the gaits of two individuals, and to give reasons why their gaits are similar and different.

Information Hiding, 2000

Many techniques for watermarking of digital images have appeared in numerous publications. Most o... more Many techniques for watermarking of digital images have appeared in numerous publications. Most of these techniques are sensitive to cropping and/or affine distortions (eg, rotation and scaling). In this paper we describe a method for the recovery of original size and appearance of images based on the concept of identification marks (“fingerprints”); the method does not require the use of the “original” image, but only a small number of salient image points. We show that, using our method, it is possible to recover original ...

Advances in information security, 2001

Springer eBooks, 2000

Frontiers in Rehabilitation Sciences

2017 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), 2017

Security and Communication Networks, 2016

2020 42nd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC)

2021 43rd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC), 2021

Information Hiding, 2000

Many techniques for watermarking of digital images have appeared in numerous publications. Most o... more Many techniques for watermarking of digital images have appeared in numerous publications. Most of these techniques are sensitive to cropping and/or affine distortions (eg, rotation and scaling). In this paper we describe a method for the recovery of original size and appearance of images based on the concept of identification marks (“fingerprints”); the method does not require the use of the “original” image, but only a small number of salient image points. We show that, using our method, it is possible to recover original ...