scholarly journals PHD Filter for Object Tracking in Road Traffic Applications Considering Varying Detectability

Sensors ◽  
2021 ◽  
Vol 21 (2) ◽  
pp. 472
Author(s):  
Olivér Törő ◽  
Tamás Bécsi ◽  
Péter Gáspár
Keyword(s):  
Sequential Monte Carlo ◽  
Road Traffic ◽  
Information Source ◽  
Extended Period ◽  
The State ◽  
Automotive Radar ◽  
Probability Hypothesis Density ◽  
Sequential Monte Carlo Method ◽  
State Dependent ◽  
Density Filter

This paper considers the object detection and tracking problem in a road traffic situation from a traffic participant’s perspective. The information source is an automotive radar which is attached to the ego vehicle. The scenario characteristics are varying object visibility due to occlusion and multiple detections of a vehicle during a scanning interval. The goal is to maintain and report the state of undetected though possibly present objects. The proposed algorithm is based on the multi-object Probability Hypothesis Density filter. Because the PHD filter has no memory, the estimate of the number of objects present can change abruptly due to erroneous detections. To reduce this effect, we model the occlusion of the object to calculate the state-dependent detection probability. Thus, the filter can maintain unnoticed but probably valid hypotheses for a more extended period. We use the sequential Monte Carlo method with clustering for implementing the filter. We distinguish between detected, undetected, and hidden particles within our framework, whose purpose is to track hidden but likely present objects. The performance of the algorithm is demonstrated using highway radar measurements.

A Novel Sequential Monte Carlo-Probability Hypothesis Density Filter for Particle Impoverishment Problem

2016 ◽  
Vol 13 (10) ◽  
pp. 6872-6877
Author(s):  
Xu Cong-An ◽  
Xu Congqi ◽  
Dong Yunlong ◽  
Xiong Wei ◽  
Chai Yong ◽  
...  
Keyword(s):  
Monte Carlo ◽  
Sequential Monte Carlo ◽  
Probability Hypothesis Density ◽  
Practical Applications ◽  
Probability Hypothesis Density Filter ◽  
Highly Nonlinear ◽  
Density Filter ◽  
Simulation Results ◽  
Application Prospects ◽  
Particle Impoverishment

As a typical implementation of the probability hypothesis density (PHD) filter, sequential Monte Carlo PHD (SMC-PHD) is widely employed in highly nonlinear systems. However, diversity loss of particles introduced by the resampling step, which can be called particle impoverishment problem, may lead to performance degradation and restrain the use of SMC-PHD filter in practical applications. In this paper, a novel SMC-PHD filter based on particle compensation is proposed to solve the problem. Firstly, based on an analysis of the particle impoverishment problem, a new particle compensatory method is developed to improve the particle diversity. Then, all the particles are integrated into the SMC-PHD filter framework. Compared with the SMC-PHD filter, simulation results demonstrate that the proposed particle compensatory SMC-PHD filter is capable of overcoming the particle impoverishment problem, which indicate good application prospects.

scholarly journals Dolphin Whistle Track Estimation Using Sequential Monte Carlo Probability Hypothesis Density Filter

2015 ◽  
Vol 62 (1) ◽  
pp. 17-20
Author(s):  
Imtiaz Ahmed
Keyword(s):  
Monte Carlo ◽  
Sequential Monte Carlo ◽  
Ground Truth ◽  
Tracking Performance ◽  
Estimation Accuracy ◽  
Common Dolphin ◽  
Stenella Coeruleoalba ◽  
Probability Hypothesis Density ◽  
Delphinus Delphis ◽  
Density Filter

This article focuses on possible automation of dolphin whistle track estimation process within the context of Multiple Target Tracking (MTT). It provides automatic whistle track estimation from raw hydrophone measurements using the Sequential Monte Carlo Probability Hypothesis Density (SMC-PHD) filter. Hydrophone measurements for three different types of species namely bottlenose dolphin (Tursiops truncates), common dolphin (Delphinus delphis) and striped dolphin (Stenella coeruleoalba) have been used to benchmark the tracking performance of the SMC-PHD filter against three major challenges- the presence of multiple whistles, spontaneous death/birth of whistles and multiple whistles crossing each other. Quantitative analysis of the whistle track estimation accuracy is not possible since there is no ground truth type track for the dolphin whistles. Hence visual inspection of estimated tracks has been used corroborate the satisfactory tracking performance in the presence of all three challenges. DOI: http://dx.doi.org/10.3329/dujs.v62i1.21954 Dhaka Univ. J. Sci. 62(1): 17-20, 2014 (January)

Tracking of Multiple Moving Objects in Dynamic Image of Omni-Directional Camera Using PHD Filter

2008 ◽  
Vol 12 (1) ◽  
pp. 16-25 ◽  
Author(s):  
Norikazu Ikoma ◽  
◽  
Ryuichi Yamaguchi ◽  
Hideaki Kawano ◽  
Hiroshi Maeda ◽  
...  
Keyword(s):  
State Space ◽  
Sequential Monte Carlo ◽  
Moving Objects ◽  
State Space Model ◽  
The State ◽  
Random Set ◽  
Probability Hypothesis Density ◽  
Space Model ◽  
Dynamic Image ◽  
Specific Object

A method of multiple moving objects tracking in dynamic image of omni-directional camera has been proposed. Finite random set (FRS) based state space model is employed in the method due to its inherent nature capable to represent the scene having occlusion and appearance of object as well as missing and false detection in observation. Sequential Monte Carlo (SMC) implementation of Probability hypothesis density (PHD) filter has been used for estimating state of the state space model. The state is a finite random set of single object states, where each element of the set consists of position and velocity of the object in panoramic image coordinate of omni-directional camera image. We propose a new method to display tracking result from weighted particles obtained from the estimation process by SMC implementation of PHD filter. Key idea of the method is to put an integer label on each particle, where the label indicates specific object among multiple objects in the image scene tracked by the particle. Numerical simulation and real image experiments illustrate tracking performance of the proposed method.

scholarly journals Implicit particle filtering for models with partial noise, and an application to geomagnetic data assimilation

2012 ◽  
Vol 19 (3) ◽  
pp. 365-382 ◽  
Author(s):  
M. Morzfeld ◽  
A. J. Chorin
Keyword(s):  
Data Assimilation ◽  
Particle Filtering ◽  
Sequential Monte Carlo ◽  
Scalar Function ◽  
The State ◽  
Stochastic Pdes ◽  
State Variables ◽  
Random Maps ◽  
Sequential Monte Carlo Method ◽  
Geomagnetic Data

Abstract. Implicit particle filtering is a sequential Monte Carlo method for data assimilation, designed to keep the number of particles manageable by focussing attention on regions of large probability. These regions are found by minimizing, for each particle, a scalar function F of the state variables. Some previous implementations of the implicit filter rely on finding the Hessians of these functions. The calculation of the Hessians can be cumbersome if the state dimension is large or if the underlying physics are such that derivatives of F are difficult to calculate, as happens in many geophysical applications, in particular in models with partial noise, i.e. with a singular state covariance matrix. Examples of models with partial noise include models where uncertain dynamic equations are supplemented by conservation laws with zero uncertainty, or with higher order (in time) stochastic partial differential equations (PDE) or with PDEs driven by spatially smooth noise processes. We make the implicit particle filter applicable to such situations by combining gradient descent minimization with random maps and show that the filter is efficient, accurate and reliable because it operates in a subspace of the state space. As an example, we consider a system of nonlinear stochastic PDEs that is of importance in geomagnetic data assimilation.

scholarly journals State Estimation and  Smoothing for the  Probability Hypothesis  Density Filter

2021 ◽  
Author(s):  
◽  
Sergio I. Hernandez
Keyword(s):  
Monte Carlo ◽  
State Estimation ◽  
Sequential Monte Carlo ◽  
Signal To Noise Ratio ◽  
Recursive Algorithm ◽  
Difficult Problem ◽  
Automated System ◽  
Order Moment ◽  
Probability Hypothesis Density ◽  
Density Filter

<p>Tracking multiple objects is a challenging problem for an automated system, with applications in many domains. Typically the system must be able to represent the posterior distribution of the state of the targets, using a recursive algorithm that takes information from noisy measurements. However, in many important cases the number of targets is also unknown, and has also to be estimated from data. The Probability Hypothesis Density (PHD) filter is an effective approach for this problem. The method uses a first-order moment approximation to develop a recursive algorithm for the optimal Bayesian filter. The PHD recursion can implemented in closed form in some restricted cases, and more generally using Sequential Monte Carlo (SMC) methods. The assumptions made in the PHD filter are appealing for computational reasons in real-time tracking implementations. These are only justifiable when the signal to noise ratio (SNR) of a single target is high enough that remediates the loss of information from the approximation. Although the original derivation of the PHD filter is based on functional expansions of belief-mass functions, it can also be developed by exploiting elementary constructions of Poisson processes. This thesis presents novel strategies for improving the Sequential Monte Carlo implementation of PHD filter using the point process approach. Firstly, we propose a post-processing state estimation step for the PHD filter, using Markov Chain Monte Carlo methods for mixture models. Secondly, we develop recursive Bayesian smoothing algorithms using the approximations of the filter backwards in time. The purpose of both strategies is to overcome the problems arising from the PHD filter assumptions. As a motivating example, we analyze the performance of the methods for the difficult problem of person tracking in crowded environments</p>

Fast Gaussian Mixture Probability Hypothesis Density Filter

2014 ◽  
Vol 568-570 ◽  
pp. 550-556
Author(s):  
Wei Hua Wu ◽  
Jing Jiang ◽  
Chong Yang Liu ◽  
Xiong Hua Fan
Keyword(s):  
Sequential Monte Carlo ◽  
Substantial Reduction ◽  
Gaussian Mixture ◽  
Computational Time ◽  
Time Cost ◽  
Probability Hypothesis Density ◽  
Relative Value ◽  
Density Filter ◽  
Computational Intractability ◽  
Practice Time

Although the Gaussian mixture probability hypothesis density (GMPHD) filter is a multi-target tracker that can alleviate the computational intractability of the optimal multi-target Bayes filter and its computational complex is lower than that of sequential Monte Carlo probability hypothesis density (SMCPHD), its computational burden can be reduced further. In the standard GMPHD filter, each observation should be matched with each component when the PHD is updated. In practice, time cost of evaluating many unlikely measurements-to-components parings is wasteful, because their contribution is very limited. As a result, a substantial reduction in complexity could be obtained by directly setting relative value associated with these parings. A fast GMPHD algorithm is proposed in the paper based on gating strategy. Simulation results show that the fast GMPHD can save computational time by 60%~70% without any degradation in performance compared with standard GMPHD.

Sign in / Sign up

Export Citation Format

Share Document