Time-varying Bias Estimation for Asynchronous Multi-sensor Multi-target Tracking Systems Using STF

Time-varying bias estimation problem was studied for multi-target tracking systems with asynchronous sensors without knowing bias dynamics. We considered general situations where the number of sensors was arbitrary as well as their sampling rates and initial sampling instants. A two-layer fusion structure was adopted. For each target, a pseudo-measurement of sensor bias was generated by fusing sensor measurements of this target. To make the pseudo-measurement decoupled from the target states the fusion coefficient matrix was determined to be a basis for the left null space of an augmented observation matrix. The bias estimation algorithm was proposed based on the Strong tracking filter （STF） by fusing pseudomeasurements. The proposed algorithm makes use of all available sensor information, has strong tracking ability to abrupt changes and avoids the matrix inversion for large dimensional matrices. Finally, the performance of the proposed algorithm is illustrated by the numerical simulation.     

Low-Complexity Adaptive Moving Target Tracking Algorithm over Multipath Fading Channels

The time-varying multipath channel and angular spreads caused by multipath and mobility of mobile stations degrade the performance of the conventional Direction-of-arrival （DOA） tracking algorithms. Moreover, although the DOA estimation methods based on the maximum likelihood have higher resolution than the beamforming and the subspace based methods, prohibitively heavy computation limits their practical applications. Hence, In this paper we propose a low-complexity adaptive moving target tracking algorithm, which includes a suboptimal DOA estimation algorithm that combines the advantages of the lower complexity of subspace algorithm and the high accuracy of ML based algorithms, and a Kalman filtering based adaptive tracking algorithm that models the dynamic property of directional changes for mobile terminals. Simulation results show that this proposed algorithm has better performance of DOA estimation and tracking of MS than the conventional ML or subspace based algorithms in terms of accuracy and robustness over multipath fading channels.     

Gaze Tracking Based Regions of Interest Detection Using a Web Camera

Regions of interest （ROIs） detection is one of the key techniques in content-based image retrieval. According to the researches on human vision in biological psychology, the interesting regions or objects can be determined by gaze frequency of human beings. The presented research addresses a novel approach of Regions of interest （ROIs） detection which is on the basis of gaze tracking using an ordinary web camera. This approach utilized an ordinary web camera to capture facial images, in which the eye regions were detected by the methods of fast face detection algorithm based on I-Iaar-like features and adaptive eye template matching. Gaze points were then estimated by the neural network whose net-input vectors consist of face position, center of pupil and glint points. Pupil center was computed based on the centroid of the pupil boundary which was obtained through morphological operations. And the glint points were located through the full-range searching in eye regions, aOIs were obtained by computing the density of gaze points. The experimental results show that the proposed method is objectively effective to measure ROIs and performs well for gaze tracking with an ordinary web camera. This approach is good applicable options for usual users and common applications, especially handicapped.     

Frequency-domain Tracking Method of GPS Signals

A tracking method of GPS signals in frequency domain is presented. The Fast Fourier transform （FFT） is used as a tool for domain conversions from time to frequency. The frequency-domain tracking method contains： C/A code tracking, carrier tracking and navigation data demodulation. In C/A code tracking process, FFT and Inverse FFT （IFFT） is used for circular correlation by which the correlation peaks of each C/A code phase are obtained. The peak points near the maximum peak are saved, for which triangle least squares approximation is applied to estimate the accurate C/A code phase. And the output of IFFT is also used for navigation data demodulation. In carrier tracking process, interpolated FFT method is applied to estimate the frequency of carrier wave. At last, simulation is performed to validate this frequency-domain algorithm.     

3D Freehand Tracking Based on Relevancy Among Local Motion Models

A novel 3D freehand tracking algorithm based on relevancy among local motion models is put forward. Firstly, a specification of the Cognitive and behavioral model （CBM） called PAMT is proposed. Secondly we regard PAMT as a data structure upon which freehand tracking algorithm is designed, and we describe the PAMT in detail. Lastly, the experimental results are provided. The proposed algorithm is tested in a virtual assembly platform and two other application systems. The highlights of this paper are as follows： （1） A new cognitive and behavioral model, called PAMT, is presented; （2） The PAMT is explained with cognitive model; （3） Focus on describing ＇Attractor in PAMT with the relevancy among local motion models; （4） Shows us how the PAMT is shaped and used to design the 3D freehand tracker. One of the advantages of PAMT and RLMM model is that it is easier to explore some of the complex correlations among the variables of the 3D hand model. Our experimental results show that, compared with the particle filter and the annealed particle filter, our algorithm effectively reduces dimensionality and can track 3D hand in real-time.     

Key Factor-Based Initialization for 3D Hand Tracking

The discovery of three-dimensional （3D） hand models corresponding to the user＇s 3D hand pose in initial frames is significant in 3D human hand tracking and interacting. This study proposes an approach to initial- ize 3D hand gestures for the user to have an easier, more pleasurable, and satisfactory experience. This paper cov- ers the following points. First, a new approach to selecting a human hand gesture from the hand postures database is presented. Second, both techniques of visualization and human-computer interaction are used in the initialization process, through which the 3D human hand model is fine- tuned time after time until the required accuracy is satis- fied. Lastly, the proposed initialization method is applied to our online virtual assembling system. We introduce a key factor to improve time cost of initialization. Our ex- perimental results show that the proposed approach is not only fast, accurate, and robust hut also direct, natural, and convenient for operators to handle.     

多目标跟踪的观测站最优机动策略

在多目标跟踪问题中,观测站的有效机动可以提高观测信息的质量,从而提升目标跟踪的精度。对此,文中提出一种基于高斯混合概率假设密度(GM-PHD)滤波器的观测站最优机动马尔可夫决策方法。首先,用Fisher信息矩阵(FIM)行列式建立代价函数;然后,计算出马尔可夫链的转移矩阵,利用马尔可夫决策过程(MDP)来获得观测站最优机动策略。其中,利用GM-PHD滤波器来估计目标的实际位置和为每一决策周期提供概率假设密度(PHD)。通过实验仿真,验证了该机动策略在提高多目标跟踪精度方面的有效性。     

基于双马尔科夫链的势概率假设密度滤波

针对已有的基于双马尔科夫链(PMC)模型的势概率假设密度(PMC-CPHD)滤波算法无法实现的问题,将PMC-CPHD算法改进为多项式形式以便于算法的实现,并给出了改进算法的高斯混合(GM)实现。实验结果表明给出的GM实现能够有效实现多目标跟踪,并且比基于PMC模型的概率假设密度(PMC-PHD)算法的GM实现提高了目标个数估计的稳定性。     

认知无线电中基于高斯混合概率假设密度滤波的主用户跟踪算法

基于随机集的高斯混合概率假设密度滤波算法是一种典型的多目标跟踪算法,可以在目标数目未知的情况下进行多目标跟踪,但是该算法要求已知目标的起始位置,在很多情况下,目标的起始位置信息是无法获得的。本文针对这一问题,提出了改进的高斯混合概率假设密度滤波算法,并将本文算法应用于认知无线电系统的主用户跟踪问题。该算法利用双向预测的方式对检测结果进行估计,即使用正向预测算法来估计现存主用户的位置,然后采用后向预测算法来搜索新生的主用户并估计出新生主用户的位置。本文算法的主要优点是在主用户的数目、出现的时间和起始位置均未知的情况下仍可以有效的跟踪目标。最后,通过仿真对本文算法的性能进行了分析。仿真结果表明,本文算法在误检率较高的情况下可以准确地跟踪主用户。      

非线性高斯系统边缘分布多目标贝叶斯滤波器

为解决存在数据关联不确定、检测不确定和杂波情况下的多目标跟踪问题,提出了一种新的多目标贝叶斯滤波器.代替维持多目标状态的联合后验密度,所提出的贝叶斯滤波器联合传递各个目标状态的边缘分布和它们的存在概率.为了处理目标运动和传感器测量模型中的非线性,利用无迹变换技术提出了一种非线性高斯条件下边缘分布贝叶斯滤波器的近似实现算法.仿真实验结果表明,与PHD(Probability Hypothesis Density)滤波器相比,所提出的滤波器具有更好的多目标跟踪能力.     

一种IMM和JPDA混合型多目标跟踪算法研究

本文研究基于IMMJPDA算法的多机动目标跟踪,并对IMMJPDA算法中聚矩阵的构成进行了改进.当目标采用不同模型时,将产生多个相互独立的聚矩阵和可行矩阵,同时得到相应的可行矩阵的条件概率.最后利用模型概率对上述条件概率进行加权求和得到关联概率.改进后的IMMJPDA算法在RMSE超调性能上有了一定的提高.     

扩展目标GMPHD滤波器的航迹维持算法

概率假设密度(PHD)滤波器提供了各目标在每个时刻的状态信息,但未形成航迹。该文提出了扩展目标高斯混合PHD(GMPHD)滤波器的航迹维持算法。首先对后验概率强度的每个高斯分量添加标签;其次在后验概率强度随着时间进行演化时,标签也随之演化;并通过航迹管理方案为扩展目标提供航迹。为提高GMPHD滤波器性能,提出了自适应的量测集划分法。最后通过仿真验证了提出算法的有效性。     

多传感器箱粒子PHD滤波多目标跟踪算法

针对目标检测概率较低导致单个传感器无法对目标进行有效检测并跟踪的问题,本文提出了多传感器箱粒子概率假设密度(multi-sensor box particle probability hypothesis density filter,MS-BOX-PHD)滤波器。MS-BOX-PHD滤波器首先将多个传感器的量测转换、融合成为一个量测集合,并利用箱粒子概率假设密度(box particle probability hypothesis density filter,BOX-PHD)滤波器对多个目标的状态进行预测和更新。数值实验表明,相较于单传感器箱粒子概率假设密度(Single-BOX-PHD)滤波器,MS-BOX-PHD滤波器在目标检测概率较低时,能够有效地对多目标的状态和数目进行估计;相较于区间量测下多传感器标准PHD粒子(multi-sensor standard probability hypothesis density particle filter with interval measurement,IM-PHD-PF)滤波器,在达到相同的跟踪性能时,计算效率提升了38.57%。     

一种有轨迹标识的利用测量生成新目标密度的GM-PHD滤波器

在存在杂波、漏检、目标数目未知和变化的情况下,PHD滤波器是一种多目标跟踪新方法,GM-PHD滤波器是PHD滤波器的一种近似实现。然而,GM-PHD滤波器没有提供单个目标状态估计的身份,而构建目标运动轨迹需要目标状态估计的身份,同时,现有的GM-PHD滤波器在新目标密度生成时对新目标出现位置进行了限制,难以对观测空间任意位置随机出现的目标进行跟踪。为解决非线性观测系统GM-PHD滤波器中目标状态估计的身份标识和新目标密度生成问题,设计了一种新的GM-PHD滤波器。该滤波器利用传感器的观测数据生成新目标密度,通过给滤波器输出的高斯项增加专有身份标识并使用身份标识将源于同一目标不同时刻的目标状态估计关联起来。仿真实验验证了滤波算法的有效性。      

A new multiple extended target tracking algorithm using PHD filter

A new multiple extended target tracking algorithm using the probability hypothesis density (PHD) filter is proposed in our study, to solve problems on tracking performance degradation of the extended target PHD (ET-PHD) filter under the nonlinear conditions and its intolerable computational requirement. It is noted that with the current Gaussian mixture implement of ET-PHD filter satisfying tracking performance could only be obtained under linear and Gaussian conditions. To extend the application of ET-PHD filter for nonlinear models, our study has derived a particle implement of ET-PHD (ET-P-PHD) filter. Our study finds that the main factors influencing the computational complexity of the ET-P-PHD filter are the partition number of measurement set and the calculation of non-negative coefficients of cells in partitions. With the pretreatment of measurements and application of a new K-means clustering based measurement set partition method, we have successfully decreased the partition number. In addition, a gating method for target state space, which is based on likelihood relationship between target state and measurement, is proposed to simplify the calculation of non-negative coefficients. Simulation results show that the algorithms proposed by our study could satisfyingly deal with multiple extended target tracking issues under nonlinear conditions, and lead to significantly lower computational complexity with tiny effect on tracking performance.     

Target Birth Intensity Estimation Using Measurement‐Driven PHD Filter

The probability hypothesis density (PHD) filter is an effective means to track multiple targets in that it avoids explicit data associations between the measurements and targets. However, the target birth intensity as a prior is assumed to be known before tracking in a traditional target‐tracking algorithm; otherwise, the performance of a conventional PHD filter will decline sharply. Aiming at this problem, a novel target birth intensity scheme and an improved measurement‐driven scheme are incorporated into the PHD filter. The target birth intensity estimation scheme, composed of both PHD pre‐filter technology and a target velocity extent method, is introduced to recursively estimate the target birth intensity by using the latest measurements at each time step. Second, based on the improved measurement‐driven scheme, the measurement set at each time step is divided into the survival target measurement set, birth target measurement set, and clutter set, and meanwhile, the survival and birth target measurement sets are used to update the survival and birth targets, respectively. Lastly, a Gaussian mixture implementation of the PHD filter is presented under a linear Gaussian model assumption. The results of numerical experiments demonstrate that the proposed approach can achieve a better performance in tracking systems with an unknown newborn target intensity.     

基于雷达测量的多目标联合检测、跟踪与分类方法

利用雷达测量中的目标速度、加速度等属性信息, 基于跳转马尔科夫系统模型高斯混合概率假设密度滤波算法, 提出了一种多目标联合检测、跟踪与分类方法.该方法在进行雷达多目标测量信息处理的多模型混合高斯概率假设密度滤波过程中, 对各高斯项编号, 进行航迹提取, 在滤波处理的同时形成带有航迹编号的明确航迹, 并进行航迹管理; 同时, 根据目标运动模型, 联合利用目标加速度控制输入与速度估计进行多目标分类.仿真试验验证了该方法能够在检测、跟踪的同时, 对目标航迹进行有效类型识别.     

一种改进的方差自适应滤波算法

标准的基于”当前”统计模型的自适应卡尔曼滤波算法中机动频率和加速度极限值存在靠经验预先设定的问题,以及在跟踪非机动和弱机动目标时存在精度不高的问题,本文在分析已有的加速度方差自适应算法的基础上,提出了一种改进的加速度方差自适应算法.仿真结果表明本文提出的改进的加速度方差自适应算法是有效性的,较已有算法提高了跟踪非机动或弱机动目标的精度.     

基于随机集的RBPF多目标关联跟踪算法

针对大量杂波环境下数量变化的纯角度多目标航迹关联跟踪问题,提出一种新的基于Rao-Blackwellized粒子采样(RBPF)航迹关联的高斯混合概率假设密度(GMPHD)滤波算法.算法首先利用GMPHD在每时刻对多个目标组成的随机集合进行估计;然后利用基于随机有限集的RBPF对GMPHD所得到的目标集合进行检测和关联...     

一种Camshift优化的粒子滤波跟踪算法

针对传统粒子滤波目标跟踪算法中用先验转移概率作分布函数时计算量大、粒子退化严重且未考虑最新观察信息等缺点,提出了一种Camshift优化的粒子滤波跟踪算法.算法首先在粒子滤波框架下,利用Camshift算法使粒子向目标状态的最大后验核密度估计方向移动.然后针对目标所处环境的不同,提出了适时调整参与Camshift算法优化的粒子数的方法,既考虑了跟踪算法的效率又考虑了粒子的多样性.跟踪结果表明,该算法的跟踪性能明显优于传统的粒子滤波算法,具有很好的实时性和鲁棒性.