Bulent Sankur

Bogazici University · Department of Electrical and Electronic Engineering

Recent advances in intelligent surveillance systems have enabled a new era of smart monitoring in a wide range of applications from health monitoring to homeland security. However, this boom in data gathering, analyzing and sharing brings in also significant privacy concerns. We propose a Compressive Sensing (CS) based data encryption that is capab...

Tensor decompositions have many application areas in several domains where one key application is revealing relational structure between multiple dimensions simultaneously and thus enabling the compression of relational data. In this paper, we propose the Discriminative Tensor Decomposition with Large Margin (shortly, Large Margin Tensor Decomposit...

Security monitoring via ubiquitous cameras and their more extended in intelligent buildings stand to gain from advances in signal processing and machine learning. While these innovative and groundbreaking applications can be considered as a boon, at the same time they raise significant privacy concerns. In fact, recent GDPR (General Data Protection...

Security monitoring via ubiquitous cameras and their more extended in intelligent buildings stand to gain from advances in signal processing and machine learning. While these innovative and ground-breaking applications can be considered as a boon, at the same time they raise significant privacy concerns. In fact, recent GDPR (General Data Protectio...

We develop a through-the-wall radar imaging (TWRI) system using stepped-frequency radar for the detection of stationary objects at close distances. This system uses the random frequency sampling and a structural sparsity based reconstruction method. The proposed reconstruction algorithm employs the block sparse structure and smoothness characterist...

Compressive Sensing (CS) theory asserts that sparse signal reconstruction is possible from a small number of linear measurements. Although CS enables low-cost linear sampling, it requires non-linear and costly reconstruction. Recent literature works show that compressive image classification is possible in CS domain without reconstruction of the si...

We consider the problem of linear data hiding or watermark embedding directly onto compressively sensed measurements (CSMs). In our encoding and decoding scheme, we seek exact recovery of concealed data and a small reconstruction error for a sparse signal under the additive noise model. We propose an efficient Alternating Direction of Methods of Mu...

In large scale distributed sensing systems such as wireless sensor networks (WSNs), Distributed Source Coding Methods can be difficult to apply, due to lack of signal statistics. Distributed Compressive Sensing (DCS) emerges as a cure to this problem. Moreover, the size of the sensor reports on distributed sensing is limited due to bandwidth and ba...

Binary descriptors have been very popular in recent years. One reason is that the algorithms that use them become computationally and memory-wise efficient. Furthermore, they tend to have some inherent robustness against some geometrical variations and against various brightness changes. These changes might result from both internal factors and ext...

SIP (Session Initiation Protocol) is one of the most common protocols that enables session control in today's communication networks. The SIP networks are also targeted by the malicious users. This study focuses on adaptive intelligent systems that detect the changes on the network flow using anomaly detection methods. Two different change point mo...

We designed SymPaD framework, a model-driven visual dictionary construction and description method, with new shape models and quantized shape library. We demonstrate that, with this new design, the most current model-driven dictionary construction method is outperformed with even smaller dictionary in object recognition and image retrieval tasks.

Experimenting with large-scale real world data is crucial for the development of network protocol and investigate their performance. However, collecting such data from real networks, and especially to annotate them with ground truth proves to, if not impossible, too tedious. In such cases use of simulated data, generated for various network scenari...

Most of the real-world signals we encounter in real-life applications have low information content. In other words, these signals can be well approximated by sparse signals in a proper basis. Compressive sensing framework uses this fact and attempts to represent signals by using far fewer measurements as compared to conventional acquisition systems...

In this paper, we propose a method for activity recognition from videos based on sparse local features and hypergraph matching. We benefit from special properties of the temporal domain in the data to derive a sequential and fast graph matching algorithm for GPUs. Traditionally, graphs and hypergraphs are frequently used to recognize complex and of...

In this paper we conduct a statistical analysis of the sensitivity and consistency behavior of objective image quality measures. We categorize the quality measures and compare them for still image compression applications. The measures have been categorized into pixel difference-based, correlation-based, edge-based, spectral-based, context-based an...

In this paper we investigate performance metrics for quantitative evaluation of object-based video segmentation algorithms. The metrics address the case when ground-truth video object planes are available. The proposed metrics are used to evaluate three essentially different approaches for video segmentation, i.e., an edge-based (1), a motion clust...

We propose a new local image descriptor named SymPaD for image understanding. SymPaD is a probability vector associated with a given image pixel and represents the attachment of the pixel to a previously designed shape repertoire. As such the approach is model-driven. The SymPad descriptor is illumination and rotation invariant, and extremely flexi...

Evaluating the performance of computer vision algorithms is classically done by reporting classification error or accuracy, if the problem at hand is the classification of an object in an image, the recognition of an activity in a video or the categorization and labeling of the image or video. If in addition the detection of an item in an image or...

In this paper, we derive two novel learning algorithms for time series clustering; namely for learning mixtures of Markov models and mixtures of hidden Markov models. Mixture models are special latent variable models that require the usage of local search heuristics such as Expectation Maximization (EM) algorithm, that can only provide locally opti...

A labeled text corpus made up of Turkish papers' titles, abstracts and keywords is collected. The corpus includes 35 number of different disciplines, and 200 documents per subject. This study presents the text corpus' collection and content. The classification performance of Term Frequcney - Inverse Document Frequency (TF-IDF) and topic probabiliti...

Graphs and hyper-graphs are frequently used to recognize complex and often non-rigid patterns in computer vision, either through graph matching or point-set matching with graphs. Most formulations resort to the minimization of a difficult energy function containing geometric or structural terms, frequently coupled with data attached terms involving...

In this study we propose a model-driven codebook generation method used to assign probability scores to pixels in order to represent underlying local shapes they reside in. In the first version of the symbol library we limited ourselves to photometric and similarity transformations applied on eight prototypical shapes of flat plateau, ramp, valley,...

In this paper, a new scaling based information hiding approach with high robustness against noise and gain attack is presented. The host signal is assumed to be stationary Gaussian with first-order autoregressive model. For data embedding, the host signal is divided into two parts, and just one patch is manipulated while the other one is kept uncha...

Excessive depth perception in 3D video is one of the major factors that causes discomfort to the viewer and that can decrease the viewer's quality perception of 3D video. With the idea of real-time quality control of 3D videos, we proposed an edge-based sparse disparity estimation algorithm with a novel similarity metric. The comparative assessment...

In this paper, we develop a graph-based method to align two dynamic skeleton sequences, and apply it to both action recognition tasks as well as to the objective quantification of the goodness of the action performance. The automated measurement of "action quality" has potential to be used to monitor action imitations, for example, during a physica...

This paper considers the watermark embedding problem onto Compressive Sensed measurements of a signal that is sparse in a proper basis. We propose a novel watermark encoding-decoding algorithm that exploits the sparsity of the signal to achieve dense watermarking. The proposed algorithm is robust under additive white Gaus-sian noise as well as impu...

In this work, we propose a novel approach for clustering Hidden Markov Models (HMMs). We use spectral learning for latent variable models to learn HMM parameters in each cluster. Unlike conventional expectation-maximization algorithms, spectral learning enables us to do parameter estimation in latent variable models without iterating, in local opti...

In this work, we propose a watermark algorithm that embeds information directly onto Compressive Sensed Measurements of a sparse signal. Proposed watermarking algorithm outperforms the classical ℓ2 and ℓ1 minimization algorithms in terms of watermark embedding capacity. Experimental results show that the proposed algorithm has a robust reconstructi...

Compressive sensing enables the reconstruction of a signal from its small number of samples in a sparse domain. It is advantageous to use compressive sensing to achieve dense signals in situations where measurements are costly, as in the case of disparity maps. In this study, disparity values are reconstructed from samples taken of the ground truth...

In this study the focus was on the one of the latest trending topics, 3D object recognition, which became trending by the developments in the 3d imaging technologies. One of the methods that is used was developed directly for 3D object recognition and the other one was developed for 2D leaf recognition. The second one was adapted for 3D object reco...

Humans carry out object recognition as a very primitive task. However, knowing all the detailed information regarding an object is mostly not possible. Therefore, object recognition and linking the object to associated data is a popular research area. In this study, we have measured leaf recognition performances of different feature and data sets u...

Face landmarking, defined as the detection and localization of certain characteristic points on the face, is an important intermediary step for many subsequent face processing operations that range from biometric recognition to the understanding of mental states. Despite its conceptual simplicity, this computer vision problem has proven extremely c...

In this study, we have experimented with different image and shape descriptors on the automatic leaf recognition problem. We have studied the effects of gross shape descriptors, Fourier descriptors, multiscale distance descriptors, and the combination of these on the leaf recognition performance using two different datasets. We have achieved 94.62%...

The paradigm of Facial Action Coding System (FACS) of-fers a comprehensive solution for facial expression measure-ments. FACS defines atomic expression components called Action Units (AUs) and describes their strength on a five-point scale. Despite considerable progress in AU detection, the AU intensity estimation has not been much investigated. We...

Graph matching is one of the principal methods to formulate the correspondence between two set of points in computer vision and pattern recognition. However, most formulations are based on the minimization of a difficult energy function which is known to be NP-hard. Traditional methods solve the minimization problem approximately. In this paper, we...

Facial Action Coding System (FACS) is the de facto standard in the analysis of facial expressions. FACS describes expressions in terms of the configuration and strength of atomic units called Action Units: AUs. FACS defines 44 AUs and each AU intensity is defined on a nonlinear scale of five grades. There has been significant progress in the litera...

We consider the problem of emotion recognition in faces as well as subject identification in the presence of emotional facial expressions. We propose alternative solutions for this identification and recognition problems using the idea of sparsity, in terms of Sparse Representation based Classifier (SRC) paradigm. In both cases, the problem is...

This paper presents the estimation of the vergence region, which is defined by the set of zero disparity points on the stereo images, in the form of the best border line separating the positive and negative disparities, by using a sparse disparity map. Sparse disparities are summed along radiating directions and the best direction to separate the s...

In this work we are presenting a sparse disparity map extraction procedure based on block matching approach. The blocks are taken around the edge locations in the reference image and searched in the target image by evaluating matching costs for each search location. For this block matching approach, the performances of cost calculation methods, suc...

In this paper, we addressed the alignment problem of two video sequences, i.e., a model sequence and a test sequence (generally longer), by graph matching on space and time domains. We applied the proposed method to a currently hot research topic, i.e., human action recognition. Space-time graphs enable a unified representation for local neighborho...

In this work, the curve compression problem is approached with a model-based probabilistic framework. We propose three different models. The proposed models can be used for purposes such as feature extraction or compression. The first model we propose is basically a Bayesian regression model for fitting piece-wise defined segments. The second model...

The problem of source separation in two dimensions is studied in this paper. The problem is formulated in the Bayesian framework. The sources are modelled as MRFs to accommodate for the spatially correlated structure of the sources, which we exploit for separation in 2D. The difficulty of working analytically with general Gibbs distributions is ove...

Automatic detection of facial expressions attracts great attention due to its potential applications in human–computer interaction as well as in human facial behavior research. Most of the research has so far been performed in 2D. However, as the limitations of 2D data are understood, expression analysis research is being pursued in 3D face modalit...

In this work, we approach the piecewise curve approximation problem with a model-based probabilistic framework. For this purpose, we propose three different models. These models can be used for feature extraction or compression. The first model is a variant of the Bayesian regression model where we can parametrically alter the design matrix. The se...

Many biometric systems, such as face, fingerprint and iris have been studied extensively for personal verification and identification purposes. Biometric identification with vein patterns is a more recent approach that uses the vast network of blood vessels underneath a person's skin. These patterns in the hands are assumed to be unique to each ind...

Automatic analysis of head gestures and facial expressions is a challenging research area and it has significant applications in human-computer interfaces. We develop a face and head gesture detector in video streams. The detector is based on face landmark paradigm in that appearance and configuration information of landmarks are used. First we det...

View-based indexing schemes for 3D object retrieval are gaining popularity since they provide good retrieval results. These schemes are coherent with the theory that humans recognize objects based on their 2D appearances. The viewbased techniques also allow users to search with various queries such as binary images, range images and even 2D sketche...

2D face classification problem is very difficult because of the illumination, pose, expression and occlusion factor. The sparse Representation based classification (SRC) provides a significant amount of robustness against to illumination changes and noise. In this study, sparse approximation is used on face recognition and applied to “Extended Yale...

Due to its potential for human-computer interfaces and human facial behavior research, automatic analysis of facial expres- sions has been an active area of study. In this paper a novel data- driven approach is proposed to detect Action Units (AUs) on 3D faces. With this approach, it is possible to design detectors that can perform detailed face re...

In this work, we propose an automatic maneuver assessment system for radio-controlled aerobatic model helicopter compe- titions. The flight videos are processed to extract the helicopter so that we can estimate its flight path. The geometrical accu- racy and the precision of the performed maneuver is measured in terms of maneouver penalty functions...

We address the problem of 2D face classification under adverse conditions. Faces are difficult to recognize since they are highly variable due to such factors as illumination, expression, pose, occlusion and resolution. We investigate the potential of a method where the face recognition problem is cast as a sparse approximation. The sparse approxim...

Automatic analysis of head gestures and facial expressions is a challenging research area and it has signiflcant applications in human- computer interfaces. In this study, facial landmark points are detected and tracked over successive video frames using a robust method based on subspace regularization, Kalman prediction and reflnement. The tra- je...

We consider two novel representations and feature extraction schemes for automatic recognition of emotion related facial expressions. In one scheme facial landmark points are tracked over successive video frames using an effective detector and tracker to extract landmark trajec-tories. Features are extracted from landmark trajectories using Indepen...

Vein pattern is the vast network of blood vessels underneath a person's skin. These patterns in the hands are assumed to be unique to each individual and they do not change over time except in size. The properties of uniqueness, stability and strong immunity to forgery of the vein patterns make it a potentially good biometric trait. In this study,...

We introduce a similarity learning scheme to improve the 3D object retrieval performance in a relevance feedback setting. The proposed algorithm relies on a score fusion approach that linearly combines elementary similarity scores originating from different shape descriptors into a final similarity function. Each elementary score is modeled in term...

We propose to model the image differentials of astrophysical source maps by Student's t-distribution and to use them in the Bayesian source separation method as priors. We introduce an efficient Markov Chain Monte Carlo (MCMC) sampling scheme to unmix the astrophysical sources and describe the derivation details. In this scheme, we use the Langevin...

In statistical shape analysis, subspace methods such as PCA, ICA and NMF are commonplace, whereas they have not been adequately investigated for indexing and retrieval of generic 3D models. The main roadblock to the wider employment of these methods seems to be their sensitivity to alignment, itself an ambiguous task in the absence of common natura...

In human facial behavioral analysis, Action Unit (AU) coding is a powerful instrument to cope with the diversity of facial expressions. Almost all of the work in the literature for facial action recognition is based on 2D camera images. Given the performance limitations in AU detection with 2D data, 3D facial surface information appears as a viable...

In this study, we analyze head gestures and facial expressions in face video streams. Facial landmark trajectories, which are the tracked coordinates of the landmarks in x and y directions, are extracted via an automatic and robust facial landmark tracking algorithm. Both raw features and features intuitively selected to reflect mimics are used. Ex...

Automatic facial action unit (AU) detection is a research topic that finds many applications in behavioral science and human computer interaction. The AU detection performance in 2D images are maturing but are not yet adequate. In this study, we develop a method to detect AUs in 3D images and show its superiority vis-a-vis 2D. The data modality is...

We handle the problem of detecting and classifying face pose views in images at the same time developing a Multi-class view detection. In order to solve this problem we use Multi-class LogitBoost algorithm in order to construct corresponding classifier structure. Although approaches generally use binary classifiers for each view class detection, we...

In this paper, we present a people finding algorithm in cluttered scenes using Histogram of Oriented Gradients (HOG) approach. The people search operation in an image is performed by sliding a detection window and converting the content of each window to a HOG feature vector. HOG feature vectors are obtained from windows with humans and not contain...

This paper investigates the use of the Compressive Sensing (CS) technique to the classification issue. In this context, CS is used as a means to probe the nonlinear manifold on which faces under various illumination effects reside. The scheme of randomly sampled faces (Randomfaces) with nearest neighbor classifier are compared with two classical fe...

In this work, we present a pose-invariant shape matching methodology for complete 3D object models. Our approach is based on first describing the objects with shape descriptors and then minimizing the distance between descriptors over an appropriate set of geometric transformations. Our chosen shape description methodology is the density-based fram...

We address the person-independent recognition problem of facial expressions using static 3D face data. The novel approach to the facial expression recognition uses non-rigid registration of surface curvature features. 3D face data is cast onto 2D feature images, which are then subjected to elastic deformations in their parametric space. Each Action...

We propose an adaptive Monte Carlo Markov Chain (MCMC) simulation for the Bayesian source separation problem and apply it to the unmixing of astrophysical components. In this method, we use the Langevin stochastic equation for transitions, which enables parallel drawing of random samples from the posterior, and which reduces the computation time si...

Automatic analysis of head and facial gestures is a significant and challenging research area for human-computer interfaces. We propose a robust face-and head gesture analyzer. The analyzer exploits trajectories of facial landmark positions during the course of the head gesture or facial expression. The trajectories themselves are obtained as the o...

In this paper, a new scaling-based image-adaptive watermarking system has been presented, which exploits human visual model for adapting the watermark data to local properties of the host image. Its improved robustness is due to embedding in the low-frequency wavelet coefficients and optimal control of its strength factor from HVS point of view. Ma...

We address content-based retrieval of complete 3D object models by a probabilistic generative description of local shape properties. The proposed shape description framework characterizes a 3D object with sampled multivariate probability density functions of its local surface features. This density-based descriptor can be efficiently computed via k...

Non-rigid registration of 3D faces finds a variety of applications in computer vision, human-computer interaction and graphics. However, automatic non-rigid registration is a non-trival problem. In this paper, we introduce an elastic deformation based method for the non-rigid registration of 3D faces that are subject to large deformations due to ex...

In this chapter, an overview of hand recognition is given first. We then describe the BioSecure Benchmarking Framework for hand modality, which is composed of open-source software, publicly available databases, and experimental protocols. For hand image-based person recognition, two methodologies are presented, namely geometry-based and appearance-...

Language learning can only advance with practice and corrective feedback. The interactive system, SignTutor, evaluates users' signing and gives multimodal feedback to help improve signing.

We investigate the source separation problem of random fields within a Bayesian framework. The Bayesian formulation enables the incorporation of prior image models in the estimation of sources. Due to the intractability of the analytical solution, we resort to numerical methods for the joint maximization of the a posteriori distribution of the unkn...

We propose a new Monte Carlo method for the astrophysical image separation problem. In this Bayesian simulation context, we used Langevin stochastic equation to generate the samples instead of the conventional random walk model. Since the samples are produced in parallel and tested pixel-by-pixel in the Metropolis-Hasting scheme, there is a signifi...

In this paper we present the results of the SHREC'09- Generic Shape Retrieval Contest. The aim of this track was to evaluate the performances of various 3D shape retrieval algorithms on the NIST generic shape benchmark. We hope that the NIST shape benchmark will provide valuable contributions to the 3D shape retrieval community. Seven groups have p...

We investigate the problem of source separation in images in the Bayesian framework using the color channel dependencies. As a case in point we consider the source separation of color images which have dependence between its components. A Markov Random Field (MRF) is used for modeling of the inter and intra-source local correlations. We resort to G...

We develop a morphosyntax-based natural language watermarking scheme. In this scheme, a text is first transformed into a syntactic tree diagram where the hierarchies and the functional dependencies are made explicit. The watermarking software then operates on the sentences in syntax tree format and executes binary changes under control of Wordnet a...

In this study 1 , we present a fully automatic TV logo iden-tification system. TV logos are detected in static regions given by time-averaged edges subjected to post-processing operations. Once the region of interest of a logo candidate is established, TV logos are recognized via their subspace fea-tures. Comparative analysis of features has indica...

In this paper we present the results of the SHREC'09-Generic Shape Retrieval Contest. The aim of this track was to evaluate the performances of various 3D shape retrieval algorithms on the NIST generic shape benchmark. We hope that the NIST shape benchmark will provide valuable contributions to the 3D shape retrieval community. Seven groups have pa...

In this paper, we focus on the reliable detection of facial fiducial points, such as eye, eyebrow and mouth corners. The proposed algorithm aims to improve automatic land-marking performance in challenging realistic face scenarios subject to pose variations, high-valence facial expressions and occlusions. We explore the potential of several feature...

The various image-processing stages in a digital camera pipeline leave telltale footprints, which can be exploited as forensic signatures. These footprints consist of pixel defects, of unevenness of the responses in the charge-coupled device sensor, black current noise, and may originate from proprietary interpolation algorithms involved in color f...

Brain functions both in an integrated and segregated manner. This study employs the concept of neural complexity to investigate these two interrelated properties of the brain. It is also known that various parts of the brain regulate the performance of the cardiovascular system and the hemisphere asymmetry which has been observed for some cognitive...

In this study, we present a 2D particle filter realization on images modelled by MRFs. The essential feature of the approach is the discretization of the posterior density and the evaluation of the MSE estimates of the image using the discretized posterior. The calculation is completely numerical rather than analytical. For simplicity, we adapted a...

In this work, we introduce the score fusion problem for 3D object retrieval. Ongoing research in 3D object retrieval shows that no single descriptor is capable of providing fine grain discrimination required by prospective 3D search engines. We present a fusion algorithm that linearly combines similarity information originating from multiple shape...

A new 3D face database that includes a rich set of expressions, systematic variation of poses and different types of occlusions is presented in this paper. This database is designed to enable research on face recognition under adverse conditions, and facial expression analysis. Particularly, the expressions consist of a judiciously selected subset...

In this work, a TV logo identification system has been developed. In preprocessing step all logos are scaled from pixel-based representation to a fixed size of macro-pixel representation. Commonly used techniques in pattern recognition, PCA (principal components analysis), NMF (non-negative matrix factorization), DCT (discrete cosine transform), an...

This paper presents an evaluation of several 3D face recognizers on the Bosphorus database which was gathered for studies on expression and pose invariant face analysis. We provide identification results of three 3D face recognition algorithms, namely generic face template based ICP approach, one-to-all ICP approach, and depth image-based Principal...

We propose a method to do constrained parameter estimation and inference from neuroimaging data using general linear model (GLM). Constrained approach precludes unrealistic hemodynamic response function (HRF) estimates to appear at the outcome of the GLM analysis. The permissible ranges of waveform parameters were determined from the study of a rep...

Accurate detection of landmark points plays an important role in many applications, such as face verification, face tracking, face expression analysis, 3D face modelling, etc. There are a lot of approaches proposed for this problem in the literature. These methods are generally based on two types of information: local texture around a given feature...