Mohamed F. Abdelkader

Silhouette-based gesture and action recognition via modeling trajectories on Riemannian shape manifolds

This paper addresses the problem of recognizing human gestures from videos using models that are built from the Riemannian geometry of shape spaces. We represent a human gesture as a temporal sequence of human poses, each characterized by a contour of the associated human silhouette. The shape of a contour is viewed as a point on the shape space of closed curves and, hence, each gesture is characterized and modeled as a trajectory on this shape space. We propose two approaches for modeling these trajectories. In the first template-based approach, we use dynamic time warping (DTW) to align the different trajectories using elastic geodesic distances on the shape space. The gesture templates are then calculated by averaging the aligned trajectories. In the second approach, we use a graphical model approach similar to an exemplar-based hidden Markov model, where we cluster the gesture shapes on the shape space, and build non-parametric statistical models to capture the variations within each cluster. We model each gesture as a Markov model of transitions between these clusters. To evaluate the proposed approaches, an extensive set of experiments was performed using two different data sets representing gesture and action recognition applications. The proposed approaches not only are successfully able to represent the shape and dynamics of the different classes for recognition, but are also robust against some errors resulting from segmentation and background subtraction.

Integrated Motion Detection and Tracking for Visual Surveillance

Visual surveillance systems have gained a lot of interest in the last few years. In this paper, we present a visual surveillance system that is based on the integration of motion detection and visual tracking to achieve better performance. Motion detection is achieved using an algorithm that combines temporal variance with background modeling methods. The tracking algorithm combines motion and appearance information into an appearance model and uses a particle filter framework for tracking the object in subsequent frames. The systems was tested on a large ground-truthed data set containing hundreds of color and FLIR image sequences. A performance evaluation for the system was performed and the average evaluation results are reported in this paper.

Real-Time Human Detection and Tracking from Mobile Vehicles

This paper presents a real-time system for detecting and tracking humans from mobile vehicles. The system integrates human detection and tracking algorithms and employs depth information obtained from a stereo vision system. Depth information is used to limit the search space of the detector. We also present a simpler and faster variant of the popular Adaboost human detector. Experimental results demonstrate that integrating depth information vastly reduces the false detection rate compared with detection without stereo. Also, we show that integrating stereo enables the detection and tracking subsystem to operate in real-time.

Activity representation using 3D shape models

We present a method for characterizing human activities using 3D deformable shape models. The motion trajectories of points extracted from objects involved in the activity are used to build models for each activity, and these models are used for classification and detection of unusual activities. The deformable models are learnt using the factorization theorem for nonrigid 3D models. We present a theory for characterizing the degree of deformation in the 3D models from a sequence of tracked observations. This degree, termed as deformation index (DI), is used as an input to the 3D model estimation process. We study the special case of ground plane activities in detail because of its importance in video surveillance applications. We present results of our activity modeling approach using videos of both high-resolution single individual activities and ground plane surveillance activities.

Collaborative compressive spectrum sensing using kronecker sparsifying basis

Spectrum sensing in wideband cognitive radio networks is challenged by several factors such as hidden primary users (PUs), overhead on network resources, and the requirement of high sampling rate. Compressive sensing has been proven effective to elevate some of these problems through efficient sampling and exploiting the underlying sparse structure of the measured frequency spectrum. In this paper, we propose an approach for collaborative compressive spectrum sensing. The proposed approach achieves improved sensing performance through utilizing Kronecker sparsifying bases to exploit the two dimensional sparse structure in the measured spectrum at different, spatially separated cognitive radios. Experimental analysis through simulation shows that the proposed scheme can substantially reduce the mean square error (MSE) of the recovered power spectrum density over conventional schemes while maintaining the use of a low-rate ADC. We also show that we can achieve dramatically lower MSE under low compression ratios using a dense measurement matrix but using Nyquist rate ADC.

Adaptive low power detection of sparse events in wireless sensor networks

Compressive Sensing (CS) has recently opened the door for efficient algorithms to solve various data gathering problems. Among these problems is sparse events detection in wireless sensor networks. In this problem, it is desirable to reduce the sensing cost by minimizing the number of sensors and the amount of data sent by each sensor. In this paper, we model the problem of sparse event detection as a compressive support recovery problem. We exploit the sparse and the binary nature of the event signal in the reconstruction algorithm using sequential compressive sensing. This provides an efficient solution to the problem, even under the assumptions of wide sensing area and high levels of noise. Simulation results show an improved performance under different compression ratios as compared to previous CS based approaches. It also shows the robustness of the proposed approach at low SNRs.

Adaptive spectrum hole detection using Sequential Compressive Sensing

Spectrum Sensing in wideband cognitive radio networks is considered one of the challenging issues facing opportunistic utilization of the frequency spectrum. Collaborative compressive sensing has been proposed as an effective technique to alleviate some of these challenges through efficient sampling that exploits the underlying sparse structure of the measured frequency spectrum. In this paper, we propose to model this problem as a compressive support recovery problem, and apply the adaptive Sequential Compressive Sensing (SCS) approach to recover spectrum holes. We propose several fusion techniques to apply the proposed approach in a collaborative manner. The experimental analysis through simulations shows that the proposed scheme can substantially increase the probability of spectrum hole detection as compared to traditional CS recovery approaches while using a very low sampling rate analog to information converter, and without requiring the knowledge of any statistical information about the environmental noise.

Power performance enhancement of underlay spectrum sharing using microstrip patch ESPAR antenna

Improving antenna arrays characteristics such as power consumption, cost and complexity has gained wide attention in the context of cognitive radio networks. One antenna array configuration that offers great potential is the electronically-steerable parasitic array radiator (ESPAR) antenna. In this paper, we propose a transmitter patch ESPAR antenna system for underlay spectrum sharing in cognitive radio networks. We transmit symbols over switchable weakly-correlated beampatterns of the antenna system so as to maximize the transmitted power to a secondary user receiver, while constraining the interference to the primary users. Our results show the superior power performance of the proposed patch ESPAR antenna system over the dipole ESPAR antennas previously used in the underlay spectrum sharing paradigm. Furthermore, we calculate the bandwidth over which the switchable patterns of the patch ESPAR antenna remains weakly correlated, and it is found to be 131.2 MHz at a center frequency of 4.902 GHz.

Exploiting Temporal Correlation of Sparse Signals in Wireless Sensor Networks

Collecting data continuously in Wireless Sensor Networks (WSNs) with limited power and bandwidth is still a challenging issue. Recently, the sparse nature of these data motivated the use of Compressive Sensing (CS) as an efficient data gathering technique. In this paper, several algorithms are proposed to effectively exploit the temporal correlation and the sparsity inherent in sensor network data over time. These algorithms combine recent advances in compressive sensing (CS) theory, data compression, and data gathering algorithms. Experimental analysis through simulation evinces that the proposed algorithms significantly reduce the power consumption by reducing the number of sent measurements for the same Normalized Mean Square Error (NMSE).