Target tracking by fusion of random measures

In this paper we propose fusion methods for tracking a single target in a sensor network. The sensors use sequential Monte Carlo (SMC) techniques to process the received measurements and obtain random measures of the unknown states. We apply standard particle filtering (SPF) and cost-reference particle filtering (CRPF) methods. For both types of filtering, the random measures contain particles drawn from the state space. Associated to the particles, the SPF has weights representing probability masses, while the CRPF has user-defined costs measuring the quality of the particles. Summaries of the random measures are sent to the fusion center which combines them into a global summary. Similarly, the fusion center may send a global summary to the individual sensors that use it for improved tracking. Through extensive simulations and comparisons with other methods, we study the performance of the proposed algorithms. This work has been supported by the National Science Foundation under Award CCF-0515246 and the Office of Naval Research under Award N00014-06-1-0012.     

Target tracking with two passive infrared non-imaging sensors

A new solution for target tracking in air space with two infrared (IR) sensors is presented. The principle of triangulation is used as a basic method for range estimation. However, when the target directions are nearly collinear relative to the baseline, this method produces unacceptable results. The problem is solved by introducing the ratio of IR energy adsorbed at the end of a baseline in a measurement vector within the extended Kalman filter type target state estimator. Also, a recursive estimator for the extinction coefficient that describes the influence of the atmosphere is designed. This combination results in a new adaptive structure for simultaneous estimation of target kinematic states and atmospheric parameters. Such a structure performs much better than the standard triangulation method, yielding acceptable results even in the case where target directions are close to the baseline. Simulation and experimental results demonstrate the feasibility and limitations of the proposed approach.     

采用高维数据聚类的目标跟踪

根据刚体各部位具有变换一致性这一特性,提出一种采用高维数据聚类的目标跟踪方法。从数学理论方面证明提出的度量方法可以应用于目标跟踪, 称其为高维数据聚类跟踪器(HDDC tracker)。该算法框架如下,首先, 采用Harris检测器对模板与跟踪区域进行特征提取;然后利用这些特征的空间信息对所提取的特征进行编组;接着计算模板特征组与跟踪区域特征组间的仿射变换阵;最后,采用高维数据聚类对这些仿射变换阵进行度量, 将那些相似仿射阵对应的跟踪区域作为跟踪目标。实验表明: HDDC tracker能够有效地跟踪具有仿射形变的目标,并且性能优于先进跟踪算法。     

Landcover classification in MRF context using Dempster-Shafer fusion for multisensor imagery.

This work deals with multisensor data fusion to obtain landcover classification. The role of feature-level fusion using the Dempster-Shafer rule and that of data-level fusion in the MRF context is studied in this paper to obtain an optimally segmented image. Subsequently, segments are validated and classification accuracy for the test data is evaluated. Two examples of data fusion of optical images and a synthetic aperture radar image are presented, each set having been acquired on different dates. Classification accuracies of the technique proposed are compared with those of some recent techniques in literature for the same image data.     

复杂场景下基于自适应特征融合的目标跟踪算法

为提升复杂场景下目标跟踪的鲁棒性,优化模型运行效率,提出一种基于自适应特征融合的相关滤波跟踪算法。该算法采用方向梯度直方图特征和卷积神经网络来对目标进行信息构建,利用特征响应的峰值旁瓣比和旁瓣值占比自适应地确定融合系数,根据融合响应来预测目标位置。为适应场景的变化,降低光照、背景和目标形变等对跟踪的影响,引入平均峰值相关能量来设计滤波器学习率调整机制,动态地进行模型更新。通过对深度特征提取网络进行轻量化设计,降低特征网络参数,提高跟踪速度。在OTB100通用数据集上进行测试,实验结果表明:文中所提算法有效降低了干扰对目标跟踪的影响,且跟踪精度、成功率和速度整体优于对比算法。     

深度特征联合表征的红外图像目标识别方法

针对红外图像目标识别问题,提出了联合卷积神经网络和联合稀疏表示的方法。卷积神经网络学习红外目标图像的深度特征,描述目标的多层次特性。不同深度特征可实现对目标不同特性的描述,因此具有良好的互补性。综合运用多层次深度特征,可为目标识别提供更为充分的信息。分类过程中,采用联合稀疏表示对待识别样本的多层次深度特征矢量进行表征,通过不同特征矢量之间的相关性约束提升整体表示精度。因此,联合稀疏表示在利用各层次深度特征的同时,充分考察了它们之间的内在关联。根据联合稀疏表示的输出结果,按照误差最小的原则判定输入样本的目标类别。实验基于中波红外（ MWIR）目标图像数据集开展,分别在原始测试样本、噪声测试样本以及少量训练样本3类条件下对提出方法进行了测试,并与4类现有红外目标识别方法进行了对比分析。实验结果表明,提出方法在设置的3类测试条件下均可以取得优势性能,表明其对于红外图像目标识别问题具有应用潜力。     

一种基于多传感器数据融合的目标跟踪算法

在制导与跟踪系统中．雷达和红外成像传感器所固有的优缺点,决定了其自身的局限性。为综合利用上述传感器信息,提高目标跟踪精度,提出了一种基于雷达和红外成像传感器数据融合的交互多模目标跟踪算法,算法首先对红外图像进行处理,然后基于上述处理结果．利用交互多模算法对雷达观测信息进行目标跟踪,最后采用分布式数据融合算法得到最终目标的跟踪结果。在有效提高跟踪精度的同时,减少了运算量。     

基于特征融合与通道感知的无人机红外目标跟踪算法

针对手工提取特征对红外目标不敏感,导致无法准确跟踪红外目标的问题,在全卷积孪生网络框架下,融合多层深度特征并结合通道感知提出了一种无人机红外目标跟踪算法。首先使用预训练网络提取目标深度特征,分别提取待跟踪目标的Conv4-1、Conv4-3、Conv5-2层特征,进而通过梯度计算选择对于目标活动和尺度变化较为敏感的特征通道参与后序的互相关操作,并通过计算模板图像和候选区域搜索图像之间的相似度获取目标响应图。最后利用平均峰值相关能量对跟踪结果进行评估并使用卡尔曼滤波对跟踪结果进行修正。在LSOTB-TIR红外目标跟踪数据集上进行了性能测试并与当前九种优秀的算法进行了对比,实验结果表明,本文算法跟踪成功率最高,能够有效应对红外目标跟踪中热交叉、干扰源等挑战,且具有较好的实时性。     

融合灰度与显著性特征的空中红外目标跟踪

针对红外目标特征简单且信息量少导致跟踪精度不高,提出一种基于灰度和显著性特征融合的核相关滤波算法用于空中红外目标跟踪。首先,在保证目标足够特征信息量的前提下对较大的目标进行不同等级压缩。然后将提取的二维灰度特征与显著性特征按页方式进行拼接扩展成三维特征,再将融合的特征用于核相关滤波。实验表明所提算法能够适应多种环境下的空中红外目标跟踪,跟踪精度和成功率典型值分别达到84.8%和63.9%,较大部分算法有很大提高,平均跟踪速度高达125 f/s,体现出了良好的实时性。因此,本文提出的算法在保证实时性的同时提高了跟踪的可靠性,具有一定的实用意义。     

激光/红外复合扫描引信目标识别算法

针对激光、红外单模引信抗环境干扰能力差、目标识别概率不高的问题,提出了一种基于关联的多元线阵激光/红外复合扫描的模糊识别算法。该算法通过建立激光/红外线阵扫描模型,对目标的顶部三维轮廓与顶部红外辐射图像进行特征提取与关联信息融合,对比真实目标特征参数,设计了基于模糊理论的目标识别算法,得出三组识别特征各自的识别概率,在此基础上对三种识别概率继续进行模糊推理融合得出目标的最终识别概率。仿真结果表明,该复合探测目标识别概率比单模识别方法有较大幅度的提高。     

基于改进粒子滤波和自适应融合的目标跟踪

针对复杂环境下视频目标跟踪精确度低的问题,提 出了一种基于混合迭代无迹粒子滤波(HI-UPF)和关联系数自 适应融合的目标跟踪算法。首先采用统计线性回归的方法对无迹变换进行优化,提出了HI- UPF,不 仅提升了滤波精度,而且有效降低了算法的时间消耗;其次基于关联系数,采用一种自适应 融合方法,实现了加性 融合和乘性融合的自适应切换,并根据关联系数提出一种改进的自适应加性融合方法。仿真 实验表明,本文方法对 于复杂条件下的目标跟踪具有较高的精度和较强的鲁棒性。     

基于多主体决策融合的结构损伤识别

为了提高航空结构的损伤识别精度和速度,提出了一种基于互信息分类器选择的多主体决策融合方法.首先获取样本,对事先选定的模式分类器进行训练测试,得到各自的混淆矩阵;然后利用基于互信息的分类器相关度指标进行最优分类器选择,得到最优分类器组合;最后利用多主体决策融合方法,对系统进行最终识别.在航空铝制加筋板上的实验表明,本文方...     

Adaptive region-based image fusion using energy evaluation model for fusion decision

A new adaptive region-based image fusion approach is proposed. To implement image segmentation, the piecewise smooth Mumford-Shah segmentation algorithm is studied and a fast and simple method is proposed to solve the energy function. Two complementary functions u ⁺ and u ⁻ of the algorithm, which are respectively looked as objects and background of the image, are extended into the whole image domain, and they are computed by linear or nonlinear diffusion. The key to the algorithm is to make optimal fusion decisions for every segmented region. For this purpose, an evaluation approach has to be given to measure the performances of the available fusion rules. Therefore an energy-based evaluation model, derived from the Total Variation principle, is proposed. By numerical experiment it has been demonstrated that despite an increase in complexity, the new approach has shown a number of advantages over previous ones, for example the ability to preserve all relevant information and remove some of side effects such as reducing contrast and sensitive to error of registration.     

Real-time speaker tracking using particle filter sensor fusion

Sensor fusion for object tracking has become an active research direction during the past few years. But how to do it in a robust and principled way is still an open problem. In this paper, we propose a new fusion framework that combines both the bottom-up and top-down approaches to probabilistically fuse multiple sensing modalities. At the lower level, individual vision and audio trackers are designed to generate effective proposals for the fuser. At the higher level, the fuser performs reliable tracking by verifying hypotheses over multiple likelihood models from multiple cues. Unlike traditional fusion algorithms, the proposed framework is a closed-loop system where the fuser and trackers coordinate their tracking information. Furthermore, to handle nonstationary situations, the proposed framework evaluates the performance of the individual trackers and dynamically updates their object states. We present a real-time speaker tracking system based on the proposed framework by fusing object contour, color and sound source location. We report robust tracking results.     

使用GMM背景减除的红外伪装人体目标快速识别算法

针对红外监视系统对伪装后的人体目标检测处理速度慢、识别准确率低的现状,提出了一种基于混合高斯模型(GMM)的背景减除(BS)快速识别算法用于红外视觉监视系统伪装人体目标检测。首先,该方法采用Mean Shift算法构造初始化背景模型后,使用编码取样原则对像素分等级取样识别前景;再利用移动目标的时间-空间相干性,使用相邻像素预估法快速、精确提取目标图像。其次,设计了基于模板的目标区域聚合新算法,有效地解决了由于伪装造成的人体目标形状特征改变而导致的误检问题。实验证明:新方法切实提高了探测识别效率,降低了人体目标误判率,为实时红外监视提供了有效手段。     

基于区域分割的序列红外图像融合算法

针对传统的基于像素点和窗口策略的融合算法对图像特征表征的失真,提出了一种基于区域分割的序列图像融合算法.首先将序列红外图像分割为3个不同的特征区域,目标区域、背景区域以及灰度区域,并将分割结果映射到可见光图像中.随后,利用多尺度几何分析工具非下采样Contourlet变换(NSCT)有效提取图像特征的特点,根据不同区域的特性在NSCT域设计不同的融合规则.对试验结果进行主观和客观的对比,结果表明:基于区域分割的序列图像融合算法不仅能够为融合图像保留更全面、丰富的背景信息,还能够更加有效、准确地提取图像中的目标特征.该算法优于传统的基于像素点和窗口规则的融合算法,是一种有效可行的图像融合算法.     

基于H∞滤波的远距离干扰下的目标跟踪算法

提出了一种加权和(WS)-H∞滤波算法实现远距 离干扰机(SOJ)环境下的目标跟踪。算法通过使用合适的传感器模型和高斯 和(GS)似然函数,充分利用了干扰信息从而提高没有量测时的跟踪精度;同时针对干扰环境 下的量测和干扰信息的统计分布不确 定的特点,采用WS-H∞滤波算法提高整个跟踪系统的鲁棒性。仿真证明,WS-H∞滤波算法 在量测噪声和干扰估计不准确时 表现出了良好的鲁棒性,其航迹连续性和跟踪精度都明显优于GS扩展卡尔曼(GS-EKF)滤波 算法,而计算复杂度却没有明显提高。     

Data fusion for ballistic targets tracking using least squares

The paper deals with the problem of state estimation of continuous-time nonlinear system using discrete-time measurements from multiple sensors. In particular, the problem of multi-radar tracking of artillery ballistic objects is considered. A batch estimator based on the iterative least squares approach is developed using simplified and accurate models of ballistic flight. The estimator is applied to process the sequences of measurements from radars tracking the same ballistic target. Estimates of the target state over time are computed and their accuracy is compared to the estimates yielded by the extended Kalman filter. Partial estimates from multiple radars are combined using track fusion approach and propagated using the 3 degree of freedom model of ballistic flight. Accuracy of target's firing point estimation is also analysed with respect to the data rates and locations of the radars with respect to the target. Practical aspects of the proposed method are also discussed.     

A Bayesian sampling approach to decision fusion using hierarchicalmodels

Data fusion and distributed detection have been studied extensively, and numerous results have been obtained in the literature. In this paper, the design of a fusion rule for distributed detection problems is re-examined, and a novel approach using Bayesian inference tools is proposed. Specifically, the decision fusion problem is reformulated using hierarchical models, and a Gibbs sampler is proposed to perform posterior probability-based fusion. Performance-wise, it is essentially identical to the optimal likelihood-based fusion rule whenever it exists. The true merit of this approach is its applicability to various complex situations, e.g., in dealing with unknown signal/noise statistics where the likelihood-based fusion rule may not be easy to obtain or may not even exist     

Probe-based automatic target recognition in infrared imagery

A probe-based approach combined with image modeling is used to recognize targets in spatially resolved, single frame, forward looking infrared (FLIR) imagery. A probe is a simple mathematical function that operates locally on pixel values and produces an output that is directly usable by an algorithm. An empirical probability density function of the probe values is obtained from a local region of the image and used to estimate the probability that a target of known shape is present. Target shape information is obtained from three-dimensional (3-D) computer-aided design (CAD) models. Knowledge of the probe values along with probe probability density functions and target shape information allows the likelihood ratio between a target hypothesis and background hypothesis to be written. The generalized likelihood ratio test is then used to accept one of the target poses or to choose the background hypothesis. We present an image model for infrared images, the resulting recognition algorithm, and experimental results on three sets of real and synthetic FLIR imagery.