The Infona portal uses cookies, i.e. strings of text saved by a browser on the user's device. The portal can access those files and use them to remember the user's data, such as their chosen settings (screen view, interface language, etc.), or their login data. By using the Infona portal the user accepts automatic saving and using this information for portal operation purposes. More information on the subject can be found in the Privacy Policy and Terms of Service. By closing this window the user confirms that they have read the information on cookie usage, and they accept the privacy policy and the way cookies are used by the portal. You can change the cookie settings in your browser.
In this paper, we propose an approach for improving the object tracking accuracy in video surveillance scenarios by estimating and compensating the occlusion introduced by static scene objects. Specifically, the scene occluder map is first estimated by analyzing the gradient of a normalized cumulative motion map from the frames of the first several minutes of a surveillance video. Then, a scene occlusion...
To verify the validity of an tracking algorithm, its performance should be evaluated and compared with other algorithms. Unlike point target, the task of extended object tracking is to estimate the object shape in addition to its kinematic state. Especially for the shape estimation performance evaluation, there is a pressing need for measuring the degree of the similarity between the estimated shape...
This paper addresses the dim target tracking problem with respect to the multi-frame detection (MFD). Generally, the MFD strategy implicitly merges the tracking stage into its framework, by which the estimated target trajectory can be simultaneously returned when a target is declared. However, due to the batch processing manner of MFD, such products, essentially, are short state sequences in time...
The OTHR (over-the-horizon radar) based tracking methods have a common prerequisite that the virtual ionospheric height, as the key ionosphere state, should be obtained either by ionosondes or external sources. However, ionosondes can not be deployed arbitrarily and external sources may not be always available, which results in the situation that the ionosphere state is unknown. This motivates us...
Networked radar systems have shown significant advances in target tracking. Reasonable power allocation strategy can sufficiently utilize the limited power resources, hence leading to the improvement of tracking performance. However, towards the existing power allocation strategies, the system configuration is only restricted to centralized architectures. Besides, practical communication requirements...
Extended target tracking is well-studied but only a few have considered Track-before-Detect (TbD). In this paper, we address short-comings of the current TbD model solution, to improve extent estimation. Experimenting with a particle filter, we demonstrate that the current solution is inaccurate. We propose a solution that improves extent estimation. Using intermediary solutions, we demonstrate the...
In this paper, we present a fusion of stereo camera and radar targets to significantly improve the tracking of pedestrians indoors especially suited for surveillance and security applications. Indoor environments pose bigger challenges for pedestrian tracking compared to outdoor environments. This makes pedestrian tracking with only a camera or radar difficult. In this work, we demonstrate that in...
Over the last decade, context has become a key source of information for tracking problems. Context inference allows refining sensor modeling and target dynamics as well as the creation of motion constraints according to the physical and operational conditions of the scenario. This work presents two example applications: indoor and inland waterway navigation where the context information is employed...
In multi-target tracking, correct models of detection statistics, namely the probability of detection and clutter rate, are required for effective multi-target state estimation. Within a multi-target filter, the detection statistics are usually assumed as known and static. Estimating the detection statistics' parameters before the execution of the filtering algorithms is not always feasible and in...
The objective of multi-target joint detection, tracking and classification (JDTC) is to simultaneously estimate the number of targets, their kinematic states, and decide on the class labels with varying target number and observation uncertainly. Good solutions require solving these problems jointly. Recently, the conditional joint decision and estimation (CJDE) algorithm based on the generalized Bayes...
The multiple model (MM) version of Gaussian mixture probability hypothesis density (GM-PHD) filter is an effective method for multiple maneuvering target tracking. However, the model set used in the MM version of GM-PHD (MM-GM-PHD) filter is the same for each target at each time step. In this paper, we present a variable structure MM-GM-PHD (VSMM-GM-PHD) filter. Different model sets at different time...
This paper presents a new approach based on extension deformation for extended object tracking (EOT). In this approach, the extension of an object is assumed to be deformed from a reference extension by moving some control points in the latter to those in the former. That is, the properties of an extension can be fully captured by the control points, given the reference extension. Thus, modeling and...
We consider a number of sensors deployed over a large geographical area for tracking a target with linear constraints on its motion dynamics which are specified by Kalman filter conditions. The state estimates from the sensors are sent over long-haul networks to a remote fusion center, where they are fused to improve the tracking accuracy. The mismatches among the sensors in incorporating the target...
Radar and sonar provide information of both range and radial velocity to unknown objects. This is accomplished by emitting a signal waveform and computing the round trip time and Doppler shift. Estimation of the round trip time and Doppler shift is usually done separately without considering the couplings between these two object related quantities. The purpose of this contribution is to first model...
In this paper, we consider clutter estimation issue under unknown, non-uniform and time-varying clutter background. First, we use finite mixture distributions (FMD) to fit the unknown clutter. As for the parameters of the FMD, we adopt the Gibbs sampler and Bayesian information criterion (BIC) to derive and evaluate clutter parameters. The final experiments show that the proposed algorithm can effectively...
Performance evaluation of tracking methods includes methods of relative and absolute performance. Absolute tracking performance is the robust end result presented to a user which determines the product solution for real world analysis. However, to achieve robust performance, the tracking method is subject to the sensor data, filtering performance, and associated models, which requires relative performance...
Consider analysis is an estimation technique that emerged in the 1960s to account for errors in system parameters while simultaneously reducing system dimensionality, and accordingly real-time computational cost, and/or guarding against issues of observability surrounding the parameters. The multitarget joint estimation problem is one whose dynamical and observational systems contain such parameter...
This paper studies the problem of simultaneous deciding on hypotheses and estimating a random parameter. We propose a joint decision and estimation (JDE) formulation, which amounts to minimizing a risk related to both decision and estimation while decision performance is also constrained within a tolerable level. The risk used in this paper is a weighted sum of estimation costs conditioned on correct...
Human target detection and tracking have great potential in military, safety, security and entertainment applications. In this paper, a complete processing procedure is proposed for human tracking in foliage-penetration environment by multistatic radar. It consists of five main steps, including clutter suppression, target detection, measurement estimation, target localization and target tracking....
Visual object-tracking is a fundamental task applied in many applications of computer vision. Many different tracking algorithms have been used ranging from point-tracking, to kernel-tracking, to silhouette-tracking based on different appearance models chosen. This paper investigates the particle filter that is used as a tracking algorithm based on the Bayesian tracking framework. The problems that...
Set the date range to filter the displayed results. You can set a starting date, ending date or both. You can enter the dates manually or choose them from the calendar.