基于动态分簇的移动目标追踪方法

针对无线传感器网络（WSN）中目标追踪的准确性低、网络能耗过高和网络生命周期短等问题,提出基于动态分簇的移动目标追踪技术。首先,构建了双层环状动态分簇的拓扑模型（TRDC）,并提出了动态分簇的更新算法;其次,在质心定位算法基础上,考虑到节点的能量,提出了基于功率级别的质心定位（CLPL）算法;最后,为了进一步减小网络的能耗,改进CLPL算法,提出了随机性定位算法。在仿真实验中,与静态簇相比,网络周期延长了22.73%;与非环状簇相比,丢失率降低了40.79%;而追踪准确性与基于接受信号强度值（RSSI）算法相差不大。所提的追踪技术能够有效保证追踪准确度,同时降低网络能耗,减小目标丢失率。     

相似背景下的运动目标自动跟综问题研究

研究视频图像中准确实现运动目标跟踪问题,要求在视频图像中找到目标确切位置,并反馈给跟踪系统.针对传统基于特征匹配的跟踪,当被跟踪物体所处环境中存在颜色、形状接近的物体时,会出现像素特征误匹配,造成运动目标跟踪错误率较高.提出基于Markov Chain Monte Carlo数据关联的运动目标跟综方法.通过建立像素概率模型,将运动日标跟踪问题公式化,运用MCMC方法对后验概率进行采样估计,避免了,模型匹配像素点的不确定性.实验证明,运动目标跟踪方法实现了在与自身相似背景下的准确跟踪,有效降低了跟踪错误率,取得了满意的效果.     

基于FPGA的动态目标跟踪系统设计

为了解决基于PC机的视频动态目标跟踪实时性瓶颈问题,设计出一种基于FPGA的动态目标跟踪系统。设计遵循图像处理金字塔模型,针对低层和中层算法简单、数据量大且存在一定并行性等特点采用FPGA硬件实现,而高层较复杂算法使用Nios Ⅱ软核进行C语言编程。整个设计采用Verilog-HDL对算法完成建模与实现,并在QUARTUS Ⅱ上进行了综合、布线等工作,最后以Altera公司的DE2开发板为硬件平台实现了整个系统。     

动态背景运动目标的快速锁定跟踪算法

为改善动态背景运动目标跟踪的精度和计算性能,提出一种Hu_Fourier特征描述子的目标识别算法和固定模板滑动置信度最佳线性逼近法的目标锁定跟踪算法。依据提出的Hu_Fourier特征描述子,快速准确地提取出识别区域运动目标的轮廓特征,结合加权距离目标识别算法,识别出图像中所有目标,计算目标的质心坐标。利用提出的固定模板滑动置信度准则,结合识别目标的质心,锁定目标,更新目标模板。依据最佳线性逼近法,确定下一帧图像的目标搜索区域。理论分析和实验结果验证了该目标锁定跟踪算法的鲁棒性和有效性。     

双边界限定下的运动目标跟踪方法

针对目标跟踪领域中复杂背景下的目标提取、目标与目标以及目标与背景之间的相互遮挡、阴影的处理、目标跟踪的实时性及鲁棒性等问题,提出一种双边界限定下的运动目标跟踪算法。利用金字塔光流法对目标关键点进行长时间无模型跟踪,并将其与检测到的前景关键点进行不重复地融合。根据融合得到的关键点,评估尺度因子及旋转因子,并利用全部前景关键点对目标中心点进行概率性表决。再利用双边界限定方法使边界上关键点对目标进行可靠描述,提升系统的鲁棒性。     

Moving object detection using region tracking

For many vision-based systems, it is important to detect a moving object automatically. The region-based motion estimation method is popular for automatic moving object detection. The region-based method has several advantages in that it is robust to noise and variations in illumination. However, there is a critical problem in that there exists an occlusion problem which is caused by the movement of the object. The occlusion problem results in an incorrect motion estimation and faulty detection of moving objects. When there are occlusion regions, the motion vector is not correctly estimated. That is, a stationary background in the occluded region can be classified as a moving object.In order to overcome this occlusion problem, a new occlusion detection algorithm is proposed. The proposed occlusion detection algorithm is motivated by the assumption that the distribution of the error histogram of the occlusion region is different from that of the nonocclusion region. The proposed algorithm uses the mean and variance values to decide whether an occlusion has occurred in the region. Therefore, the proposed occlusion detection and motion estimation scheme detects the moving regions and estimates the new motion vector, while avoiding misdetection caused by the occlusion problem. The experimental results for several video sequences demonstrate the robustness of the proposed approach to the occlusion problem.This work was presented in part at the 8th International Symposium on Artificial Life and Robotics, Oita, Japan, January 24–26, 2003     

Distributed joint sensor registration and target tracking via sensor network

This paper deals with distributed registration of a sensor network for target tracking in the presence of false and/or missed measurements. Each sensor acquires measurements of the target position in local coordinates, having no knowledge about the relative positions (referred to as drift parameters) of its neighboring nodes. A distributed Bernoulli filter is run over the network to compute in each node a local posterior target density. Then a suitable cost function, expressing the discrepancy between the local posteriors in terms of averaged Kullback–Leibler divergence, is minimized with respect to the drift parameters for sensor registration purposes. In this way, a computationally feasible optimization approach for joint sensor registration and target tracking is devised. Finally, the effectiveness of the proposed approach is demonstrated through simulation experiments on both tree networks and networks with cycles, as well as with both linear and nonlinear sensors.     

基于Mean Shift和粒子滤波的运动目标跟踪

针对Mean Shift算法跟踪效果不佳以及粒子滤波算法计算量大且实时性不强等问题,提出了一种结合Mean Shift和粒子滤波的运动目标跟踪融合算法。首先用MeanShift算法进行跟踪,在跟踪结果不佳的情况下用粒子滤波算法进行修正。实验结果表明,融合算法很好地结合了两种算法的优点,既保留了Mean Shift算法的实时性,又很好地体现了粒子滤波算法的鲁棒性,实用性很强。     

基于视野分界线的多摄像机运动目标跟踪算法

提出了一种改进的基于视野分界线的多摄像机运动目标跟踪算法。该算法通过采用SIFT算法来自动获取特征点,生成视野分界线。当目标穿过分界线后,将待确定目标颜色直方图与目标颜色直方图相匹配,匹配度最好的赋予标识,即完成目标的交接,用Mean Shift算法完成后续目标的跟踪。在搭建的实验平台上进行了实时跟踪,实验结果表明该算法能有效实现多摄机间的目标连续跟踪。     

基于改进Camshift的穿墙雷达运动人体目标成像跟踪算法

针对穿墙雷达运动人体目标图像"闪烁"与"抖动"的特点,提出基于改进Camshift的穿墙雷达运动人体目标成像跟踪算法。首先,针对形成的连续多帧穿墙雷达图像以及对应的颜色概率分布图,引入目标预测过程以确定图像中运动目标搜索波门,消除波门外的杂波干扰;然后,利用颜色概率分布图,在波门内自适应迭代调整目标搜索窗尺度,匹配形状与大小变化的目标图像以提取目标位置;最后,对提取的目标位置进行α-β滤波,形成连续平滑的目标运动跟踪航迹,实现基于穿墙雷达成像的建筑物内运动人体稳定航迹跟踪。多输入多输出（MIMO）穿墙雷达实验结果显示,与传统Camshift和Meanshift算法相比,改进后算法的跟踪航迹误差分别降低了40.99%和43.09%,获得了更加准确和平滑的目标运动航迹。     

基于多视角航拍配准的运动小目标检测与跟踪

针对存在3D场景遮挡的航拍视频运动小目标跟踪问题,提出一种基于多视角航拍配准的运动小目标检测和跟踪算法。该算法首先对图像序列间隔采样,利用Harris检测器提取全局特征点,通过Delaunay三角网对待配准图像实现初始匹配,然后利用整合变换模型计算差分图像,并利用累积能量检测出目标,最后通过卡尔曼运动滤波消除运动目标跟踪的抖动。实验结果表明,该算法对城市和郊区场景的航拍视频可以检测出最小30个像素的缓慢运动目标。     

Narrowband target tracking using a biomimetic sonarhead

This work presents a target tracking method that exploits both the physics of narrowband echolocation and appropriate movements of the sensor, taking inspiration from narrowband (CF-FM) bats. The platform used in this work is RoBat, a biomimetic sonarhead mounted on a mobile platform. Two different techniques are integrated in the tracker. First, Barshan’s and Kuc’s method for estimating target range and azimuth angle is adapted to the geometry of the sonarhead. Next, Walker’s arc scanning hypothesis for target elevation estimation in CF-FM bats is implemented, applying curve fitting for an accurate estimation of the peak-delay. The integration of both techniques in a preliminary 3D tracking system is performed applying both methods independently.     

近邻传播观测聚类的多扩展目标跟踪算法

由于传感器分辨率高或目标存在多个反射源等原因,一个目标可以同时产生多个观测数据,对于解决这种扩展目标的跟踪问题,概率假设密度(PHD)滤波算法是一种有效的方法.针对扩展目标概率假设密度滤波算法中观测集合划分,提出一种利用近邻传播聚类方法进行观测集合划分的多扩展目标跟踪算法.实验结果表明,所提出的方法不但能够获得正确的划分观测集合,而且计算复杂度较已有划分方法有较大降低,同时在多目标跟踪效果方面优于已有算法.     

逐次分块差分直方图对运动目标的提取

为从视频图像序列中准确且完整地提取运动目标,提出一种基于运动差分直方图的目标提取算法。通过对视频图像序列中两帧相邻图像以及它们的差分图像进行逐次分块,分析对应分块的差分直方图,利用背景区域块与目标区域块对应的差分直方图分布不同,逐步去除背景提取目标,有效避免提取的目标内部出现空洞,以块为单位的处理提高算法运行速度。通过数值实验验证了该算法具有较好检测性。     

基于SVM输出概率和后置滤波的运动目标分类*

提出了一种新的运动目标分类方法,该方法利用sigmoid函数将标准SVM的输出结果直接转换为目标所属类别的概率,避免了分类器的组合问题;同时该方法还利用后置加权均值滤波器对SVM的初始输出结果进行滤波平滑处理,进一步提高了分类的正确率。实验结果表明,该方法能有效地提高运动目标分类的精度。     

均值漂移框架下基于背景差分的运动目标跟踪

提出了一种基于背景差分法原理的均值漂移MS跟踪算法。使用距离度量函数判断目标是否失去跟踪,当MS跟踪目标位置发生较大偏移时,通过使用背景差分法提取的目标形心位置对其进行修正。实验结果表明,该方法应用于实时运动目标的跟踪时具有良好的跟踪效果。     

基于阈值判断的CamShift目标跟踪算法

针对CamShift算法只利用目标的颜色信息,在跟踪过程中,易受目标相似物、遮挡以及光照等复杂背景影响导致目标搜索窗口发散,跟踪稳定性能降低,提出了一种基于阈值判断的目标跟踪方法。该方法将OTSU法和Snake模型结合,利用OTSU法以最佳阈值对图像进行分割,分离前景区域和背景区域,初步提取目标轮廓作为Snake模型的初始轮廓,经收敛得到目标的精准轮廓,利用轮廓外接最小矩形框内的像素计算目标质心,判断与CamShift算法中目标搜索窗口质心之间的欧式距离,如果未超出阈值,则直接使用CamShift算法跟踪目标,反之,则将计算出的目标质心作为CamShift算法中当前帧目标搜索窗口的质心跟踪目标。实验结果表明,该算法跟踪目标具有较好的实时性,跟踪性能稳定、可靠。     

Bearings only single-sensor target tracking using Gaussian mixtures

This paper presents a new approach for single sensor tracking using passive bearings only measurements. Gaussian mixture measurement presentation, together with a track splitting algorithm, allow space-time integration of the target position uncertainty with a simple algorithm. The bearings-only measurements are incorporated into track as they arrive using a dynamic bank of linear Kalman filters. While this approach is applicable to the case with the target detection, data association and multitarget issues, this paper concentrates on the target trajectory estimation using associated measurements. A simulation study demonstrates the benefits of this approach.     

Joint spatial registration and multi-target tracking using an extended PM-CPHD filter

An extended product multi-sensor cardinalized probability hypothesis density (PM-CPHD) filter for spatial registration and multi-target tracking (MTT) is proposed. The number and states of targets and the biases of sensors are jointly estimated by this method without the data association. Monte Carlo (MC) simulation results show that the proposed method (i) outperforms, although computationally more expensive than, the extended multi-sensor PHD filter which has been proposed for joint spatial registration and MTT; (ii) outperforms the multi-sensor joint probabilistic data association (MSJPDA) filter which is also extended in this study for joint spatial registration and MTT when the clutter is relatively dense.     

基于差分法的均值漂移单目标跟踪

研究了一种基于差分法原理的MS跟踪算法。当MS跟踪目标位置发生较大偏移时,通过使用差分法提取的目标形心位置对其进行修正。实验结果表明,该方法应用于运动目标的跟踪具有良好的跟踪效果。