Bahador Khaleghi

Multisensor data fusion: A review of the state-of-the-art

There has been an ever-increasing interest in multi-disciplinary research on multisensor data fusion technology, driven by its versatility and diverse areas of application. Therefore, there seems to be a real need for an analytical review of recent developments in the data fusion domain. This paper proposes a comprehensive review of the data fusion state of the art, exploring its conceptualizations, benefits, and challenging aspects, as well as existing methodologies. In addition, several future directions of research in the data fusion community are highlighted and described.

Random finite set theoretic based soft/hard data fusion with application for target tracking

This paper reports on ongoing research on development of data fusion systems capable of processing soft as well as hard data. Such fusion systems are distinguished from the conventional systems where input data are assumed to be provided by typically well-characterized electronic sensor systems. The incorporation of soft human-generated data into fusion process is an emerging trend in fusion community majorly motivated by asymmetric warfare situations where observational opportunities for traditional hard sensors is restricted. Random finite set theory is a mathematical framework with powerful representational and computational abilities making it a promising approach to address several fundamental challenges in soft/hard fusion systems. In this paper the first prototype soft/hard fusion system based on random finite set theory is described. Experimental results obtained using the developed system prove the plausibility as well as efficiency of a random finite set theoretic approach to fusion of soft/hard data.

Multisensor data fusion: Antecedents and directions

The notion of combining redundant and complementary sensory data to achieve higher quality information has been around for long time. For example, biological systems usually rely on fusion of multimodality sensory data to perceive the surrounding environment. There has been an ever-increasing interest in multi-disciplinary research on multisensor data fusion technology driven by its versatility and diverse areas of application. This paper presents an overview of data fusion state of the art discussing advances in fusion methodologies and architectures. In addition, future directions of research in data fusion community including large-scale, hybrid, secure fusion, and importance of standard evaluation framework are highlighted and described.

Attention Assist: A High-Level Information Fusion Framework for Situation and Threat Assessment in Vehicular Ad Hoc Networks

Driver inattentiveness constitutes the main cause of road accidents, which makes it a major factor in road safety. In this paper, we propose a comprehensive framework to address the road safety problem by tackling it from a high-level information fusion standpoint, considering vehicular ad hoc networks (VANETs) as the deployment platform. The proposed framework relies on the multientity Bayesian networks (MEBNs), which exploit the expressiveness of first-order logic for semantic relations, and the strength of the Bayesian networks in handling uncertainty. First, the entities that influence the inattention phenomenon, as well as both their causal and semantic relationships, are identified. Next, an MEBN-based high-level information fusion framework is proposed through which entities, situations, and their relationships in specific contexts are modeled using MEBN fragments. Furthermore, MEBN inference is used to assess the situations of interest by estimating their states. To demonstrate the capabilities of the proposed framework, a collision warning system simulator has been developed, which evaluates the likelihood of a vehicle being in a near-collision situation using a wide variety of local and global information sources available in various VANET environments. If the threat of being in a near-collision situation is determined to be high, then the driver is warned accordingly. Our experimental results for two distinct single-vehicle and multivehicle categories of driving scenarios, as well as a novel hybrid MEBN inference, demonstrate the capability of the proposed framework to efficiently achieve situation and threat assessment on the road.

Human-centered multi-target tracking: A Random Set theoretic approach

The incorporation of soft human-generated data into the fusion process is an emerging trend in the data fusion community. This paper describes an extension of our original Random Set (RS) theoretic soft/hard data fusion system from single-target to multi-target tracking case. Leveraging recent developments in the RS theoretic data fusion community, we propose a novel soft measurement-to-track association algorithm. Based on this algorithm, we describe a multi-target tracking system capable of processing soft human-generated data. Our preliminary experiments demonstrate the advantages of the proposed soft data association algorithm (SDAA) in achieving substantial improvement of tracking performance, considering the baseline algorithm to rely merely on human opinions for solving the data association problem.

Distributed random set theoretic soft-hard data fusion: Target tracking application

The development of data fusion systems capable of incorporating soft human-generated data into the fusion process is an emerging trend in the fusion community, motivated mainly by asymmetric warfare situations where the observational opportunities for traditional hard sensors are restricted. This paper describes an extension of our prototype soft/hard data fusion system, based on the random set theory, from centralized into a fully distributed computational framework. A fully distributed data fusion algorithm relies only on information exchange between local sensor nodes and hence promises enhanced scalability, reliability, and robustness in contrast to the conventional centralized fusion approach. We propose a novel approach for distributed estimation of average soft data using the consensus propagation algorithm. The distributed estimation of aggregated hard data is accomplished through an average consensus filter. Based on the proposed approach, we describe a single-target tracking system capable of processing soft and hard data. The preliminary experiments demonstrate the efficiency of the consensus propagation based approach for distributed aggregation of soft data, as well as the advantages of incorporating soft data into the distributed data fusion process.

SILT: Scale-invariant line transform

Line matching is useful in many computer vision tasks such as object recognition, image registration, and 3D reconstruction. The literature on line matching has advanced in recent years, nevertheless, compared to other features (such as point and region matching approaches) it has made little progress. Especially, very few algorithms address the problem of image scaling. In this paper, we present a new line detection and matching algorithm that is invariant to image scale variation (SILT). The algorithm detects line segments as local extrema in the scale-space. Each detected line segment is represented in a distinctive manner using Haar-like features. PCA is further deployed to improve upon the compactness and robustness of representation. Experimental results demonstrate the effectiveness of the proposed approach to deal with image scale variations.

Random set theoretic soft/hard data fusion framework

Relying on the random set theory, a new data fusion framework is proposed to enable fusion of both soft and hard data. First, soft and hard data are modeled in a unified approach. Next, a novel method for soft data trustworthiness estimation is developed. The experimental results demonstrate significant performance improvement achieved through systematic incorporation of soft data into the data fusion process.