Tracking multiple moving targets with swarms of mobile robots

This paper presents a distributed approach to enable mobile robot swarms to track multiple targets moving unpredictably. The proposed approach consists of two constituent algorithms: local interaction and target tracking. When the robots are faster than the targets, Lyapunov theory can be applied to show that the robots converge asymptotically to each vertex of the desired equilateral triangular configurations while tracking the targets. Toward practical implementation of the algorithms, it is important to realize the observation capability of individual robots in an inexpensive and efficient way. A new proximity sensor that we call dual rotating infrared (DRIr) sensor is developed to meet these requirements. Both our simulation and experimental results employing the proposed algorithms and DRIr sensors confirm that the proposed distributed multi-target tracking method for a swarm of robots is effective and easy to implement.     

Detection Avoidance and Priority-Aware Target Tracking for UAV Group Reconnaissance Operations

In this paper, we present a novel approach for stationary target tracking in reconnaissance operations with a small UAV group. A reconnaissance mission has multiple competing requirements, such as short scan time and repetitive scanning of the entire area, target recognition, and target tracking. Especially in real-world military reconnaissance scenarios, different types of targets with hostile characteristics exist. The UAVs must scan and track the targets while avoiding detection by enemies. Although small UAVs are unlikely to be detected, they become prone to detection if their path is predictable. To meet these competitive requirements, we propose an attractive pheromone-based cooperative path planning method that makes path prediction almost impossible by ensuring a random path selection mechanism. To avoid detection during target tracking, we implement a new discrete-time tracking scheme with random time intervals and random path planning for multiple UAVs. The proposed model enables a UAV group to sporadically scan the entire area, quickly locate the targets, and simultaneously track the targets based on their priority. In addition, it offers a mechanism that permits the command and control center to balance between reconnaissance and target tracking operations to meet every mission requirement.     

强跟踪输入估计概率假设密度多机动目标跟踪算法

针对多机动目标跟踪中,目标数目未知及加速度不确定的问题,提出一种强跟踪输入估计（modifiedinputestimation,MIE）概率假设密度多机动目标跟踪算法．在详细分析算法的基础上,通过引入强跟踪多重渐消因子,以不同速率实时调节滤波器各个通道的预测协方差及相应的滤波器增益,从而实现MIE算法对加速度未知或发生人幅度突变的机动目标白适应跟踪能力;并将该算法与概率假设密度滤波算法有效结合,町以较好地跟踪未知数目的多机动目标．仿真结果表明,新算法比传统的多机动目标跟踪算法具有更岛的跟踪精度,且具有较好的实时性．     

Adaptive swarm optimization for locating and tracking multiple targets

Locating and tracking multiple targets in the dynamic and uncertain environment is a crucial and challenging problem in many practical applications. The main task of this paper is to investigate three fundamental problems, which are composed of the identification of irregular target, locating multiple targets and tracking multiple targets. Firstly, the proposed objective function successfully gets the target's shape to discern eccentric target in the specific environment. Secondly, for the sake of locating multiple targets, the adaptive PSO algorithm divides the swarm into many subgroups, and adaptively adjusts the number of particles in each subgroup by the competition and cooperation technology. Thirdly, in order to track multiple targets in the dynamic environment, the proposed swarm optimization has the characteristic of the adaptively covered radius of the subgroup according to the minimum distance among other subgroups. To show the efficiency and high performance of the proposed algorithms, several algorithms chiefly concentrate on locating and tracking three ants in the practical systems.     

Improving the localization accuracy of targets by using their spatial–temporal relationships in wireless sensor networks

Due to the low cost and capabilities of sensors, wireless sensor networks (WSNs) are promising for military and civilian surveillance of people and vehicles. One important aspect of surveillance is target localization. A location can be estimated by collecting and analyzing sensing data on signal strength, time of arrival, time difference of arrival, or angle of arrival. However, this data is subject to measurement noise and is sensitive to environmental conditions, so its location estimates can be inaccurate. In this paper, we add a novel process to further improve the localization accuracy after the initial location estimates are obtained from some existing algorithm. Our idea is to exploit the consistency of the spatial–temporal relationships of the targets we track. Spatial relationships are the relative target locations in a group and temporal relationships are the locations of a target at different times. We first develop algorithms that improve location estimates using spatial and temporal relationships of targets separately, and then together. We prove mathematically that our methods improve the localization accuracy. Furthermore, we relax the condition that targets should strictly keep their relative positions in the group and also show that perfect time synchronization is not required. Simulations were also conducted to test the algorithms. They used initial target location estimates from existing signal-strength and time-of-arrival algorithms and implemented our own algorithms. The results confirmed improved localization accuracy, especially in the combined algorithms. Since our algorithms use the features of targets and not the underlying WSNs, they can be built on any localization algorithm whose results are not satisfactory.     

OATS: Oxford Aerial Tracking System

Small robot helicopters are becoming a popular research platform due to the availability of off-the-shelf components and their suitability for useful applications. We describe the Oxford Aerial Tracking System (OATS) that we are commissioning, which takes a commercial airframe and low-level flight controller, and adapts these for use in applications requiring the visual tracking of ground targets. This uses a camera on a two-axis gimbal feeding images into an on-board processing system, which communicates summary information to a ground station. So far we have tested the system off the aircraft using laboratory targets and canned image sequences; we have also developed in simulation a variety of high-level algorithms for the platform, including those for target area scanning, geo-localisation, 3-D path planning, and target trajectory prediction; this paper focuses on the hardware, system software, and object tracking sub-systems. Testing on the airframe is now underway.     

基于不同运动模型的IMMPDA算法仿真

在现有的机动目标跟踪算法中,其中的概率数据关联（PDA）算法和交互式多模型（IMM）算法最具代表性。而在此基础上发展而来的IMMPDA算法是解决杂波环境下单机动目标跟踪问题比较有效的方法。通过对分别基于CA模型、Singer模型和“当前”统计模型的交互式多模型概率数据关联（IMMPDA）算法进行仿真,对其优缺点进行对比分析。仿真结果显示IMMPDA算法在高机动目标跟踪中具有巨大优势,不同的：运动模型基于IMMPDA都较好地实现了对高速高机动目标的滤波跟踪。     

面向无人机的视觉目标跟踪算法：综述与展望

近年来,无人机因其小巧灵活、智能自主等特点被广泛应用于民用和军事等领域中,特别是搜索侦察过程中首要的目标跟踪任务。无人机视觉目标跟踪场景的复杂性和运动目标的多变性,使得目标特征提取及模型建立困难,对目标跟踪性能带来巨大的挑战。本文首先介绍了无人机视觉目标跟踪的研究现状,梳理了经典和最新的目标跟踪算法,特别是基于相关滤波的跟踪算法和基于深度学习的跟踪算法,并对比了不同算法的优缺点。其次,归纳了常用的目标跟踪数据集和性能评价指标。最后,展望了无人机视觉目标跟踪算法的未来发展趋势。     

融合图像感知哈希技术的运动目标跟踪

目的 为解决运动目标跟踪时因遮挡、尺度变换等产生的目标丢失以及传统匹配跟踪算法计算复杂度高等问题,提出一种融合图像感知哈希技术的运动目标跟踪算法.方法 本文算法利用感知哈希技术提取目标摘要进行模板图像识别匹配,采用匹配跟踪策略和搜索跟踪策略相配合来准确跟踪目标,并构建模板评价函数和模板更新准则实现目标模板的自适应更新,保证其在目标发生遮挡和尺度变换情况下的适应性.结果 该算法与基于NCC(normalized cross correlation)的模板匹配跟踪算法、Mean-shift跟踪算法以及压缩跟踪算法相比,在目标尺度变换和物体遮挡时,跟踪的连续性和稳定性更好,且具有较低的计算复杂度,能分别降低跟踪系统约6.2%、 6.3%、 9.3%的计算时间.结论 本文算法能有效实现视频场景中目标发生遮挡及尺度变换情况下的跟踪,跟踪的连续性和稳定性良好,且算法具有较低的计算复杂度,有利于实时性跟踪系统的构建.     

Designing energy efficient target tracking protocol with quality monitoring in wireless sensor networks

Target tracking is one application of wireless sensor networks and energy efficient target tracking algorithms that can be used for accurate tracking are highly desired. In order to achieve energy savings, we focus on reducing energy usage by limiting the number of sensors used to track a target through monitoring their data quality and by limiting the amount of data being sent to the cluster head. We propose an energy efficient target tracking protocol that uses two algorithms to accomplish this goal. Simulation studies show that the network lifetime is extended up to 35% with application of both algorithms and that the side effect on target tracking accuracy is not too negative.     

深度学习辅助的多行人跟踪算法

目的 目标的长距离跟踪一直是视频监控中最具挑战性的任务之一。现有的目标跟踪方法在存在遮挡、目标消失再出现等情况下往往会丢失目标,无法进行持续有效的跟踪。一方面目标消失后再次出现时,将其作为新的目标进行跟踪的做法显然不符合实际需求;另一方面,在跟踪过程中当相似的目标出现时,也很容易误导跟踪器把该相似对象当成跟踪目标,从而导致跟踪失败。为此,提出一种基于目标识别辅助的跟踪算法来解决这个问题。方法 将跟踪问题转化为寻找帧间检测到的目标之间对应关系问题,从而在目标消失再现后,采用深度学习网络实现有效的轨迹恢复,改善长距离跟踪效果,并在一定程度上避免相似目标的干扰。结果 通过在标准数据集上与同类算法进行对比实验,本文算法在目标受到遮挡、交叉运动、消失再现的情况下能够有效地恢复其跟踪轨迹,改善跟踪效果,从而可以对多个目标进行持续有效的跟踪。结论 本文创新性地提出了一种结合基于深度学习的目标识别辅助的跟踪算法,实验结果证明了该方法对遮挡重现后的目标能够有效的恢复跟踪轨迹,适用在监控视频中对多个目标进行持续跟踪。     

近年目标跟踪算法短评——相关滤波与深度学习

目的 目标跟踪是计算机视觉领域的重要组成部分。近年来,基于相关滤波和深度学习的目标跟踪算法层出不穷,本文拟对经典的若干目标跟踪算法进行阐述与分析。方法 首先,对基于相关滤波跟踪算法的基础理论进行介绍,针对相关滤波算法在特征改进类、尺度改进类、消除边界效应类、图像分块类与目标响应自适应类方面进行总结;接下来,从3个方面对基于深度学习的目标跟踪算法进行阐述与分析：目标分类、结构化回归、孪生网络,并对有代表性的跟踪算法的优势与缺陷进行较深层次的解读。结果 通过列举跟踪算法在相关滤波阶段、深度学习阶段针对不同的改进机制的改进算法,总结各阶段算法的优缺点。对目标跟踪算法的最新进展进行阐述,最终对目标跟踪算法的未来发展方向进行总结。结论 基于相关滤波的目标算法在实时性方面表现优秀,但对于复杂背景干扰、相似物遮挡等情况仍然需要优化。深层的卷积特征对于目标有强大的表示力,通过使相关滤波算法与深度学习结合,大幅度提升了算法表现力。基于深度学习的跟踪算法则更侧重于跟踪的性能,大多无法满足实时性。孪生神经网络的使用对于基于深度学习类目标跟踪算法产生了很大的推动,兼顾了算法的性能和实时性。     

Improved assignment with ant colony optimization for multi-target tracking

Detecting and tracking ground targets is crucial in military intelligence in battlefield surveillance. Once targets have been detected, the system used can proceed to track them where tracking can be done using Ground Moving Target Indicator (GMTI) type indicators that can observe objects moving in the area of interest. However, when targets move close to each other in formation as a convoy, then the problem of assigning measurements to targets has to be addressed first, as it is an important step in target tracking. With the increasing computational power, it became possible to use more complex association logic in tracking algorithms. Although its optimal solution can be proved to be an NP hard problem, the multidimensional assignment enjoyed a renewed interest mostly due to Lagrangian relaxation approaches to its solution. Recently, it has been reported that randomized heuristic approaches surpassed the performance of Lagrangian relaxation algorithm especially in dense problems. In this paper, impelled from the success of randomized heuristic methods, we investigate a different stochastic approach, namely, the biologically inspired ant colony optimization to solve the NP hard multidimensional assignment problem for tracking multiple ground targets.     

Adaptive multi-modality sensor scheduling for detection and tracking of smart targets

This paper considers the problem of sensor scheduling for the purposes of detection and tracking of “smart” targets. Smart targets are targets that can detect when they are under surveillance and react in a manner that makes future surveillance more difficult. We take a reinforcement learning approach to adaptively schedule a multi-modality sensor so as to most quickly and effectively detect the presence of smart targets and track them as they travel through a surveillance region. An optimal scheduling strategy, which would simultaneously address the issue of target detection and tracking, is very challenging computationally. To avoid this difficulty, we use a two stage approach where targets are first detected and then handed off to a tracking algorithm. We investigate algorithms capable of choosing whether to use the active or passive mode of an agile sensor. The active mode is easily detected by the target, which makes the target prefer to move into hide mode. The passive mode is nearly undetectable to the target. However, the active mode has substantially better detection and tracking capabilities then the passive mode. Using this setup, we characterize the advantage of a non-myopic policy with respect to myopic and random polices for multitarget detection and tracking.     

基于GLMB滤波和Gibbs采样的多扩展目标有限混合建模与跟踪算法

本文针对杂波条件下多扩展目标的状态估计, 目标个数估计, 扩展目标形状估计问题, 提出了一种基于标签随机有限集(Labelled random finite sets, L-RFS)框架下多扩展目标跟踪学习算法, 该学习算法主要包括两方面:多扩展目标动态建模和多扩展目标的跟踪估计.首先, 结合广义标签多伯努利滤波器(Generalized labelled multi-Bernoulli, GLMB)建立了扩展目标的量测有限混合模型(Finite mixture models, FMM), 利用Gibbs采样和贝叶斯信息准则(Bayesian information criterion, BIC)准则推导出有限混合模型的参数来对多扩展目标形状进行学习, 然后采用等效量测方法来替代扩展目标产生的量测, 对扩展目标形状采用椭圆逼近建模, 实现扩展目标形状与状态的估计.仿真实验表明本文所给的方法能够有效跟踪多扩展目标, 并且在目标个数估计方面优于CBMeMBer算法.此外, 与标签多伯努利滤波(LMB)计算比较表明: GLMB和LMB算法滤波估计精度接近, 二者精度高于CBMeMBer算法.     

一种基于目标分割的实时跟踪方法

为减少颜色相似背景在目标跟踪过程中对跟踪结果的影响,提出一种基于目标分割的实时跟踪方法。利用背景差分的方法进行目标分割,使用目标位置、大小及颜色的特征完成目标状态的更新,对连续帧之间的目标实施匹配跟踪,当2个目标重合分离时采用目标大小、颜色特征实现匹配。实验结果表明,该方法匹配跟踪速度较快、跟踪效果较好。     

基于多尺度建议框的目标跟踪误差修正方法

随着深度学习技术引入视觉目标跟踪领域,目标跟踪算法的精度和鲁棒性有了很大的提高。但在低空无人机跟踪目标的实际场景中,情况比较复杂,如相机的抖动、大量的遮挡、视角和焦距的改变等,使得跟踪算法的准确性受到极大挑战。目前的算法大多建立在目标外观变化缓慢的前提假设下,在跟踪的过程中不具备检测和修复漂移（跟踪误差）的能力。针对该问题,提出了一种基于多尺度建议框的目标跟踪误差修正方法。离线阶段,利用大量的已标注的目标样本训练基于多尺度建议框的目标跟踪修正模型,获取不同类别目标的先验知识。在线阶段在核相关滤波跟踪的基础上,依据相关响应置信度自适应评价的结果,通过目标跟踪修正模型不定期重新初始化目标的位置,避免了因为误差累积而导致跟踪失败。算法在无人机航拍数据集上进行了测试,结果表明,该跟踪算法在目标发生较大形变的情况下能较好的修正跟踪漂移问题。相比于其他几种算法,目标跟踪的成功率和精度分别提高了14.3%和3.1%。     

无人机多机动目标自主探测与跟踪

无人机机载相机图像中机动目标尺寸较小而且会发生显著变化,加上大量的背景噪声干扰,给目标探测和跟踪带来很大困难.针对这些问题,本文提出了一种在无人机机载相机图像序列中自主探测与跟踪多个机动目标的方法.首先,提取目标的图像数字特征并采用级联分类算法进行特征分类,得到目标的强分类器,对目标进行自主探测搜索.然后,基于全局最优关联算法对探测回波进行关联滤波,实现对多个机动目标的跟踪与识别,其中最优关联代价矩阵融合了距离和方向信息,提高了关联和跟踪的鲁棒性.将无人机航拍图像序列中的地面坦克作为目标进行实验,结果表明本文算法可以实现对多个机动目标的自主探测和跟踪,并具有较好的跟踪鲁棒性.     

基于多分辨率方法的主动轮廓线跟踪算法

由Ｋａｓｓ等１９８７年首次提出的主动次轮廓线模型，在数字图像分析和计算机视觉领域得到了越来越广泛的应用，基于单分辨率的主动轮廓线跟踪算法在一个共同的缺点，即：要求初始轮廓线离目标的真实轮廓线很近，这样的算法用于跟踪将只能跟踪缓慢运动的目标，从而限制了主动轮廓线跟踪算法的应用范围，本文提出了一种基于多分辨率的主动轮廓线跟踪算法，它不但继承了基于单分辨率的主动轮廓线算法的优点，而且降低了对初始轮廓线的     

Deep Learning and Preference Learning for Object Tracking: A Combined Approach

Object tracking is one of the most important processes for object recognition in the field of computer vision. The aim is to find accurately a target object in every frame of a video sequence. In this paper we propose a combination technique of two algorithms well-known among machine learning practitioners. Firstly, we propose a deep learning approach to automatically extract the features that will be used to represent the original images. Deep learning has been successfully applied in different computer vision applications. Secondly, object tracking can be seen as a ranking problem, since the regions of an image can be ranked according to their level of overlapping with the target object (ground truth in each video frame). During object tracking, the target position and size can change, so the algorithms have to propose several candidate regions in which the target can be found. We propose to use a preference learning approach to build a ranking function which will be used to select the bounding box that ranks higher, i.e., that will likely enclose the target object. The experimental results obtained by our method, called \( DPL ^{2}\) (Deep and Preference Learning), are competitive with respect to other algorithms.