Adaptive Auxiliary Particle Filter for Track-Before-Detect With Multiple Targets

The performance of ballistic target interception is critically dependent on the performance of the target state estimation. The estimation performance then strongly depends on the accuracy of the measurement model. The Gaussian uncertainty distribution has commonly been used for representing the statistical properties of sensor noise, due to its mathematical simplicity and effectiveness. However, seeker sensor measurements are often corrupted by glint noise which is highly non-Gaussian, and conventional Gaussian filtering algorithms are known to show unsatisfactory performance in the presence of glint noise. This research proposes the use of a particle filter for ballistic target tracking in a glint noise environment. The target tracking performance of the particle filter is compared with that of the extended Kalman filter.

Kalman filter and its families have played an important role in information gathering, such as target tracking. Data association techniques have also been developed to allow the Kalman filter to track multiple targets simultaneously. This paper revisits the principle and applications of the Kalman filter for single target tracking and multiple hypothesis tracking (MHT) for multiple target tracking. We present the brief review of the Bayes filter family and introduce a brief derivation of the Kalman filter and MHT. We show examples for both single and multiple targets tracking in simulation to illustrate the efficacy of Kalman filter-based algorithms in target tracking scenarios.

In this paper, we address the problem of detecting the appearance time of a target and tracking its state with a particle filter in the Track-Before-Detect context. We show that it is possible to model the problem as a quickest detection change problem in a Bayesian framework. In this case, the posterior density of the target time appearance is a mixture where each component represents the hypothesis that the target arrived at a given time. As the posterior density is intractable in practice, we propose to approximate each component of the mixture by a particle filter, and we show that the weights of the mixture can be computed recursively thanks to quantities provided by the different particle filters. The overall filter yields good performance.

Particle filters (or PFs) are widely used for the tracking problem in dynamic systems. Despite their remarkable tracking performance and flexibility, PFs require intensive computation and communication, which are strictly constrained in wireless sensor networks (or WSNs). Thus, distributed particle filters (or DPFs) have been studied to distribute the computational workload onto multiple nodes while minimizing the communication among them. However, weight normalization and resampling in generic PFs cause significant challenges in the distributed implementation. Few existing efforts on DPF could be implemented in a completely distributed manner. In this paper, we design a completely distributed particle filter (or CDPF) for target tracking in sensor networks, and further improve it with neighborhood estimation toward minimizing the communication cost. First, we describe the particle maintenance and propagation mechanism, by which particles are maintained on different sensor nodes and propagated along the target trajectory. Then, we design the CDPF algorithm by adjusting the order of PFs' four steps and leveraging the data aggregation during particle propagation. Finally, we develop a neighborhood estimation method to replace the measurement broadcasting and the calculation of likelihood functions. With this approximate estimation, the communication cost of DPFs can be minimized. Our experimental evaluations show that although CDPF incurs about 50% more estimation error than semi-distributed particle filter (or SDPF), its communication cost is lower than that of SDPF by as much as 90%.

In this work, we propose a new ground moving target indicator (GMTI) radar based ground vehicle tracking method which exploits domain knowledge. Multiple state models are considered and a Monte-Carlo sampling based algorithm is preferred due to the manoeuvring of the ground vehicle and the non-linearity of the GMTI measurement model. Unlike the commonly used algorithms such as the interacting multiple model particle filter (IMMPF) and bootstrap multiple model particle filter (BS-MMPF), we propose a new algorithm integrating the more efficient auxiliary particle filter (APF) into a Bayesian framework. Moreover, since the movement of the ground vehicle is likely to be constrained by the road, this information is taken as the domain knowledge and applied together with the tracking algorithm for improving the tracking performance. Simulations are presented to show the advantages of both the new algorithm and incorporation of the road information by evaluating the root mean square error (RMSE).

Particle filtering (PF) and dynamic programming (DP) are two famous track-before-detect (TBD) algorithms which have been used widely for solving problems of dim radar target detection and tracking under low signal-to-noise ratio (SNR). A new method that combines these two algorithms, via. the name DPPF, is proposed in this paper. It uses Halton sequence for generating uniformly distributed particles in the state space. After that, weights of these particles are calculated and then the rearranged weights will be integrated scans of data as the evaluation function in DP approach. This new method diversifies the particles and decreases the computational complexity due to its avoidance of predicting and updating of particles. Experimental results from both simulated and real data verify that DPPF algorithm can improve the performance of detecting and tracking dim radar target.

With the rapid development of information technology in inland waterway, radar was widely used to track the ship running, in order to ensure the safe and efficient shipping industry development, as well as to avoid a ship collision causing unnecessary economic losses. For the problem of radar target tracking data associated, a method of data association with radar data was proposed, which was based on radar target course and speed. By this method, the target error or loss problem was solved when the target density was large and used belief rule-based (BRB) methodology to verify the reliability of this method. Real target points obtained by this method had been filtered through particle filter. And then the radar target tracking was achieved. This method has a good target tracking results, verified by Monte Carlo simulations.

This paper presents a learning particle filter (LPF) to solve the problems of uncertainty, varying number, overlap, ambiguous, non-rigid, nonlinear, and non-Gaussian in tracking multiple visual targets. Evolution learning obtains a detector that guides proposal distribution originally, and then online learning renders the detector adaptive to pose altering and proposal distribution closer to posterior distribution. Simulation results demonstrate the performance of LPF algorithm.

In this paper, we present an easy-hardware-implementation multiple model particle filter (MMPF) for maneuvering target tracking. In the proposed filter, the sampling importance resampling (SIR) filter typically used for nonlinear and/or non-Gaussian application is extended to incorporating multiple models that are composed of a constant velocity (CV) model and a "current" statistical (CS) model, and the Independent Metropolis Hasting (IMH) sampler is utilized for the resampling unit in each model. Compared with the bootstrap MMPF, the proposed MMPF requires no knowledge of models and model transition probabilities for different maneuvering motions, and keeps a constant number of particles per model at all times. This allows a regular pipelined hardware structure and can be implemented in hardware easily. Furthermore, using the IMH sampler for the resampling unit avoids the bottleneck introduced by the traditional systematic resampler and reduces the latency of the whole implementation. Simulation results indicate that the proposed filter has approximately equal tracking performance with the bootstrap MMPF. Hardware architecture of the IMH sampler and its corresponding sample unit are presented, and a parallel architecture consisting of CV model processing element (PE), CS model PE and a central unit (CU) is described. The proposed architecture is evaluated on a Xilinx Virtex-II Pro FPGA platform for a maneuvering target tracking application and the results show many advantages of the proposed MMPF over existing approaches in terms of efficiency, lower latency, and easy hardware implementation.

This paper uses several case studies to examine the capabilities of the Particle Filter (PF) for multi-object tracking and classification. The first study treats the relatively simple problem of tracking two objects in the absence of any background clutter. The scenario considers a thrusting ballistic missile that expends a lower stage after burnout. The objective is to track both stages using a single PF which contains separate states for the two objects. The second case study involves a more complicated scenario in which a maneuvering air vehicle deploys a series of decoys to confuse the tracking sensor. The problem is made more complex by the fact that tracking occurs in moderately dense clutter background. The PF state vector includes a discrete binary existence state which denotes whether the decoy is present or absent. In addition to measurements of line-of-sight angles, range, etc., the filter is assumed to have available “classification” which are judgments made by the observer as to the origin of each measurement. The classifications are specified in terms of probabilities. The possible objects under track can be the target alone, a centroid of the target and decoy in the case of unresolved measurements, or separate, resolved tracks of both target and decoy.

In the application of particle filter algorithm for target tracking in wireless sensor networks, an Auxiliary particle filter (APF) and Gaussian particle filter (GPF) are discussed to solve the particle degradation problem of Particle Filter (PF). By introducing an auxiliary variable and two rounds weighted processes, APF makes the particle weights stable and relaxes particle degradation. GPF uses a Gaussian density function to approximately estimate posterior probability distribution. It doesn’t require re-sampling, and there is no particle degradation. Therefore, three algorithm are applied to single-target tracking in wireless sensor networks. Finally, the comparison of three algorithm’s performance in target tracking is presented and the simulation results are also given. From these results we can see that APF has better accuracy than PF algorithm and GPF has better accuracy and real-time performance than PF algorithm.

Detecting and tracking fiducial points successfully can generate necessary dynamic and deformable information for facial image interpretation tasks with numerous potential applications. In this paper we propose an automatic fiducial points tracking method using multiple Differential Evolution Markov Chain (DE-MC) particle filters with kernel correlation techniques. Fiducial points are initialized through the scale invariant feature based detectors. By taking the advantage of the ability to approximate complicated proposal distributions, multiple DE-MC particle filters are applied for fiducial points tracking by building a path connecting sampling with measurements, based on the fact that the posteriori depends on both the previous state and the current observation. A Kernel correlation analysis approach is proposed to find the detection likelihood with maximization of the similarity criterion between the target points and the candidate points. Sampling efficiency is improved and computational time is substantially reduced by making use of the intermediate results obtained in particle allocation.

In this paper, we present a new approach for online joint detection and tracking for multiple targets. We combine a deterministic clustering algorithm for target detection with a sequential Monte Carlo method for multiple target tracking. The proposed approach continuously monitors the appearance and disappearance of a set of regions of interest for target detection within the surveillance region. No computational effort for target tracking is expended unless these regions of interest are persistently detected. In addition, we also integrate a very efficient 2D data assignment algorithm into the sampling method for the data association problem. The proposed approach is applicable to nonlinear and nonGaussian models for the target dynamics and measurement likelihood. Computer simulations demonstrate that the proposed hybrid approach is robust in performing joint detection and tracking for multiple targets even though the environment is hostile in terms of high clutter density and low target detection probability

Underwater acoustic sensor networks (UASNs) play an important role in the ocean's protection. They can realize real-time data collection, monitoring, exploration, and many other underwater applications by connecting and coordinating seafloor sensors and underwater vehicles. To achieve these application objectives, such as fishes tracking in biological monitoring field and submarines tracking in military field, target tracking is one of the key techniques. This paper presents a centralized fusion algorithm based on the interacting multiple models and the adaptive Kalman filter (IMMCFAKF) for target tracking in UASNs. Specifically, by introducing an adaptive forgetting factor into the optimal centralized fusion Kalman filter algorithm, the optimal centralized fusion adaptive Kalman filter (CFAKF) algorithm is obtained first. Then, combining the superiorities of both the optimal CFAKF algorithm and the conventional IMM algorithm, the optimal IMMCFAKF is achieved. The numerical simulations are provided to demonstrate the effectiveness of the proposed optimal IMMCFAKF algorithm.

This work documents our investigation of multiple target tracking filters in proximity sensor networks when the target power levels are not known. The challenge is that when the targets are close, it is hard to determine if the sensor reports are the results of a loud target or multiple quiet targets. Given the binary measurements:1 for detection of targets and 0 for nondetection of targets, the works studies the feasibility of using the joint multitarget particle filter to estimate the number of targets, and at the same time estimate the 5D target states including the positions, velocities and power levels.