Search

This paper develops three-dimensional (3D) Cartesian tracking algorithms for a high-resolution wide field of view (FOV) camera surveillance system. This system consists of a network linking multiple narrow FOV cameras side-by-side looking at adjacent areas. In such a multi-camera system, a target usually appears in the FOV of one camera first, and then shifts to an adjacent one. The tracking algorithms estimate target 3D positions and velocities dynamically using the angular information (azimuth and elevation) provided by multiple cameras. The target state (consisting of position and velocity) is not fully observable when it is detected by the first camera only. Once it moves into the FOV of the next camera, the state can then be fully estimated. The main challenge is how to transfer the state information from the first camera to the next one when the target moves across cameras. In this paper, we develop an approach, designated as Cartesian state estimation with full maximum likelihood information transfer (fMLIT), to cope with this challenge. Since the fMLIT consists of an implicit state relationship, the conventional Kalman-like filters (for explicit constraints) are not suitable. We then develop three Gauss-Helmert filters, and test them with simulation data. [ABSTRACT FROM AUTHOR]

In passive sensor target tracking, there are applications that require converting the angle-only measurements into Cartesian space. Multisensor methods can be used to convert the raw measurements into Cartesian measurements by finding the intersection of lines of sight. This method contains significant nonlinearity in its conversion; therefore, it is subject to corresponding errors such as a conversion bias and an improper covariance. The proposedmethod uses a secondorder Taylor series expansion to accurately account for the conversion nonlinearities. This results in an explicit (noniterative) expression of the Cartesian position based on two line-of-sight measurements in three dimensions. This paper investigates the severity of the conversion biases from nonlinearity and the efficiency of the unbiased conversion with regard to compensating for them. [ABSTRACT FROM AUTHOR]

Detection and tracking of divers have become an important factor in port protection against underwater intruders. A problem arises from divers with open breathing systems because detections of the air bubbles they produce can mislead the tracking filter and sometimes result in a lost track. In this paper a probabilistic model is developed which reflects the probability that a false measurement originates from the bubbles. The novel contribution of this paper is the integration of this model in the probabilistic data association filter (PDAF) to improve the track continuity. The bubble detections may also cause confusion in the track initiation. To prevent this problem, a clustering method is proposed based on morphological operators which allows tracks to be initialized based on two-point differencing of the cluster centroids from succeeding scans. This morphological clustering method is included in a cell averaging constant false alarm rate (CA-CFAR) detector in such a way that both the point detections and their corresponding clusters can be fed to the tracking filter. These techniques are implemented and applied to real data of two divers, one with an open breathing system and the other with a closed breathing system, operating simultaneously in a coastal area. The real data were recorded from an active 90 kHz narrowband multibeam imaging sonar.

In multisensor tracking systems that operate in a centralized or distributed information processing architecture, measurements from the same target obtained by different sensors can arrive at the processing center out of sequence due to system latencies. In order to avoid either a delay in the output or the need for reordering and reprocessing entire sequences of measurements, such latent measurements have to be processed by the tracking filter as out-of-sequence measurements (OOSM). Recent work developed a 'one-step' procedure for incorporating OOSM with multiple-time-step latency into the tracking filter, which, while suboptimal, was shown to yield results very close to those obtained by reordering and reprocessing an entire sequence of measurements. The counterpart of this problem is the need to remove (revocate) measurements that have already been used to update a track state. This can happen in real-world systems when such measurements are reassigned to another track. Similarly to the problem of update with an OOSM, it is desired to carry out the removal of an earlier measurement without recomputing the track estimate (and the data association) using possibly a long sequence of subsequent measurements one at a time. A one-step algorithm is presented for this problem of removing a multistep OOSM.

When the size of targets is comparable to the range resolution of monopulse radars, these targets should be considered as extended rather than point targets. If several closely-spaced targets fall within the same radar beam and between adjacent matched filter samples in range, the full monopulse information from all of these samples can and should be used to resolve these targets, i.e., estimate the number of targets and their respective angles-of-arrival and ranges. To detect and localize multiple unresolved extended targets, we establish a model for monopulse radar returns from extended objects, and present a maximum likelihood estimator (MLE) to localize the targets. Rissanen's minimum description length (MDL) criterion will be used to decide the number of existing extended objects. We also derive the upper limit on the number of targets and their scattering centers that can be resolved, and we provide necessary conditions for these targets to be uniquely identified. We compare the new extended target monopulse processing scheme with previously developed point-target monopulse techniques in the simulations.

Successful track-to-track data association in a multisensor, muititarget scenario is predicated on a proper cost function. The cost function for associating two tracks is well established. This paper investigates different versions of the likelihood that more than two tracks represent the same target. These likelihoods lead to different cost functions for the general track-to-track association problem. Two of the likelihoods are approximations based on ad-hoc extensions of the well-known two-track expression. The next approximation is the generalized likelihood (GL), where the true state is replaced by the maximum likelihood (ML) estimate of the target state in the true likelihood. The final likelihood is derived exactly from a diffuse prior of the target state and is referred to as the diffuse prior likelihood (DPL). This paper reveals the connection between the DPL and GL. The last two likelihood expressions incorporate the cross-track error covariance matrices, which are not readily available in distributed track fusion. Therefore, the paper considers three forms of the DPL or GL: 1) the correlated tracks (CT) form by using the actual cross-covariance matrices, 2) the independent tracks (IT) form by assuming the cross-covariance matrices are zero, and 3) the approximated correlated tracks (ACT) form by approximating the cross-covariance matrices. Simulations compare the performance of all likelihood versions.

Recent advances are described in detecting and estimating the location of more than one target within a single monopulse radar beam. A quickest detection procedure based on sequential processing of the likelihood ratios is used for deciding on a change in the target model as an indication of a newly spawning target.

In this paper the problem of tracking multiple spawning targets with multiple finite-resolution sensors is considered and a new algorithm for measurement-to-track association with possibly unresolved measurements is presented. The goal is to initialize new tracks of spawned targets before they are resolved from the mother platform so that one has the ability to carry out early discrimination when they become resolved. The multiple scan data association problem is first formulated as a multidimensional assignment problem with explicit new constraints for the unresolved measurements. Then the top M hypotheses tracking (TMHT) is presented where the state estimates and their covariances are modified based on the M best hypotheses through the assignment solutions. A modification to the assignment problem is developed that leads to a linear programming (LP) where the optimal solution can be a noninteger in [0, 1]. The fractional optimal solution is interpreted as (pseudo) probabilities over the N - 1 frame sliding window. The best hard (binary) decision assignment solution and the M best via TMHT are compared with the soft decision solution for 2-D tracking scenarios with various sensor configurations. Based on the simulation results, the soft assignment approach has better track maintenance capability than the single best hard assignment and a performance nearly as good as the TMHT. Its computational load is slightly higher than the single best hard assignment but much lighter than TMHT.

We examine large-sample properties of the maximum-likelihood estimator (MLE) in the vicinity of points where the Fisher information measure (FIM) equals zero. Under mild regularity conditions the MLE is asymptotically efficient and therefore lower hounded by the Cramer-Rao lower hound (CRLB) [5], which diverges for such points. When a linear sensor array is used for angle-of-arrival (AOA) estimation, the CRLB diverges as the AOA approaches [tau]/2. We provide new results characterizing the MLE performance in the AOA problem.