基于车载视频监控的乘客检测及跟踪算法

针对公交车上乘客相互遮挡及光照变化明显的问题,提出一种基于头肩部边缘特征和局部不变特征的人体检测及跟踪算法。首先对待检测图像进行自适应阈值背景差分,实现乘客目标分割;然后用样本的梯度方向直方图(HOG)特征训练支持向量机(SVM)基础分类器,结合自适应增强(AdaBoost)算法提炼出最终的强分类器,对前景图像进行扫描实现乘客目标检测;最后提取目标区域和当前搜索区域的快速鲁棒性特征(SURF),通过特征点匹配实现乘客目标跟踪。实验表明,在乘客相互遮挡及光照变化明显的情况下,该算法仍具有高于80%的检测率和跟踪率,且满足系统实时性的要求,可用于客流计数。     

一种结合多特征的前方车辆检测与跟踪方法

车辆检测是汽车防碰撞预警的前提,为了提高前方车辆检测的实时性和鲁棒性,提出一种结合多特征的前方车辆检测跟踪方法。该方法不依赖车道检测,利用车底部阴影的梯度特征确定可能存在车辆的区域,使用差分盒子维计算对应区域的分形维数来排除噪声,根据车辆的水平边缘特征信息精确定位,通过卡尔曼滤波器跟踪检测到的目标,利用归一化转动惯量做车辆验证。实验结果表明,该方法能够在多种交通环境中实时有效地检测前方车辆。     

基于半监督自动谱聚类算法的网络故障检测

针对网络故障检测中利用先验知识不足和多数谱聚类算法需事先确定聚类数的问题,提出一种新的基于成对约束信息传播与自动确定聚类数相结合的半监督自动谱聚类算法。通过学习一种新的相似性测度函数来满足约束条件,改进NJW聚类算法,对非规范化的Laplacian矩阵特征向量进行自动谱聚类,从而提高聚类性能。在UCI标准数据集和网络实测数据上的实验表明,该算法较相关比对算法聚类准确率更高,可满足网络故障检测的实际需要。     

基于TLD改进的自动人体检测与实时跟踪算法

首先提出一种运动人体检测算法,通过图像序列识别出运动人体作为跟踪目标,然后在TLD算法中引入目标轨迹预测,利用该信息来辅助空间搜索。运动人体检测算法首先采用背景减除和边缘检测算法获得完整目标轮廓,然后使用HU仿射不变矩检测出运动人体,该特征能适应目标旋转、尺度、仿射等变化场合。为提高跟踪实时性,在TLD框架中引入卡尔曼轨迹预测,并先在预测位置邻域搜索。实验结果表明,该运动人体检测算法能够在静态和动态背景下准确地检测出运动人体;改进后的TLD算法与原始算法相比,在准确率不降低情况下,降低了计算复杂度。     

基于轨迹的足球检测与跟踪

在足球比赛视频中对足球进行检测和跟踪,如果球与球场线融合或球与球员形成连通区域,则得到球的位置信息将变得非常困难.针对这类问题提出了一种基于轨迹的解决方案.基于维特比算法检测并跟踪球,应用最小二乘法得到球运动轨迹的插值函数,基于所得到的插值函数,剔除维特比算法检测与跟踪球过程中的一些误检,同时补充漏检处的球位置信息.通过实验验证,算法具有较高的准确性.     

二进制无线传感器网络目标定位跟踪算法

二进制无线传感器网络是一种新的和具有巨大潜在应用前景的目标定位跟踪平台,它具有通信负荷少,生存能力强等特点。提出一种基于网络时间的迭代算法,在目标通过一定数量的节点后,给出较为精确的目标运动参数,并在算法迭代的过程中,不断优化算法结果。最后,运用多Agent仿真思想,借助于RePAST仿真平台对算法进行了仿真验证。     

Real-time multiple vehicle detection and tracking from a moving vehicle

Abstract. A real-time vision system has been developed that analyzes color videos taken from a forward-looking video camera in a car driving on a highway. The system uses a combination of color, edge, and motion information to recognize and track the road boundaries, lane markings and other vehicles on the road. Cars are recognized by matching templates that are cropped from the input data online and by detecting highway scene features and evaluating how they relate to each other. Cars are also detected by temporal differencing and by tracking motion parameters that are typical for cars. The system recognizes and tracks road boundaries and lane markings using a recursive least-squares filter. Experimental results demonstrate robust, real-time car detection and tracking over thousands of image frames. The data includes video taken under difficult visibility conditions. Received: 1 September 1998 / Accepted: 22 February 2000     

基于卡尔曼滤波器的抗遮挡车辆跟踪算法

针对由于车辆相互遮挡而不能准确跟踪车辆的问题,提出一种基于卡尔曼滤波器的跟踪算法。算法利用基于图像内容敏感度的背景提取算法,获取目标车辆的二值化图像,然后采用卡尔曼滤波器预测目标车辆在下一帧中的状态,在预测区域中进行搜索匹配,建立相邻帧中目标车辆的对应关系,有效推理目标车辆的驶入、驶出以及遮挡和分离,并根据预测信息对发生遮挡的目标进行跟踪。实验结果表明,该算法能够获得良好的跟踪效果。     

Path tracking controller design of four wheel independent steering automatic guided vehicle

This paper presents a new type of four wheel independent steering automatic guided vehicle (4WIS-AGV) for carrying heavy baggage and proposes a controller for the 4WIS-AGV to track reference trajectories. To do this task, the followings are done. Firstly, a 4WIS-AGV is designed and manufactured for experimental purpose. Secondly, a kinematic modeling for the 4WIS-AGV is introduced based on a single track vehicle model. Thirdly, based on the modeling, a controller is designed based on Backstepping method for the 4WIS-AGV to track reference trajectories. Fourthly, a control system is developed using industrial PC and AVR ATmega128 microcontrollers to implement the designed controller. Finally, simulations and experiments are conducted to verify the effectiveness and performances of the proposed controller in tracking two types of reference trajectories: a trajectory with sharp edges for parallel steering maneuver and a circular trajectory for zero-sideslip maneuver. The results show that the proposed controller can make the 4WIS-AGV track the trajectory with sharp edges and the circular trajectory very well.     

基于UWB的智能跟随车导航定位算法研究

针对目前市场上现有智能跟随车定位精度不足,提出一种基于UWB信号的定位算法。在智能跟随车的上方安置两个固定基站,手持标签到两个基站的距离数据经过卡尔曼滤波算法的处理,利用三角函数进行计算,得出标签到两个基站中点的距离和偏移角度,将距离和角度数据传送给电机控制模块,通过PID控制算法调节PWM值,从而控制电机的转速和转向。实验表明,该方法能够实现标签定位的距离误差小于9 cm,角度误差小于10°,使智能跟随车的定位更为精准。     

Model based vehicle detection and tracking for autonomous urban driving

Situational awareness is crucial for autonomous driving in urban environments. This paper describes the moving vehicle detection and tracking module that we developed for our autonomous driving robot Junior. The robot won second place in the Urban Grand Challenge, an autonomous driving race organized by the U.S. Government in 2007. The module provides reliable detection and tracking of moving vehicles from a high-speed moving platform using laser range finders. Our approach models both dynamic and geometric properties of the tracked vehicles and estimates them using a single Bayes filter per vehicle. We present the notion of motion evidence, which allows us to overcome the low signal-to-noise ratio that arises during rapid detection of moving vehicles in noisy urban environments. Furthermore, we show how to build consistent and efficient 2D representations out of 3D range data and how to detect poorly visible black vehicles. Experimental validation includes the most challenging conditions presented at the Urban Grand Challenge as well as other urban settings.     

Visual vehicle detection and tracking based on the 'sign pattern'

Real-time vehicle detection and tracking applicable for outdoor mobile robots on a structured road is proposed. The method is based on a locomotion strategy known as 'sign pattern based stereotyped motion'. The sign pattern of a vehicle is the shadow underneath. The robot detects a vehicle from a distance of 30 m with a domestic video camera and extracts its front shape in real-time processing. The distance between the robot and vehicle, and the velocity and width of the vehicle are estimated during the tracking process. In addition, the center of the vehicle in two-dimensional imaging is determined during the tracking process to estimate the type of motion. The accuracy rate of vehicle width measurement based on the shadow underneath is better than 90% in practice, which can be adequately used for vehicle avoidance applications. The applied method has been tested on roads covered with different types of shadows as well as under different weather conditions. The results of the experiments verify the validity of the method.     

基于视频的高速公路车辆检测和跟踪算法

采用一种基于像素的新方法来更新背景，同时将工作区域分为车辆检测和车辆跟踪两个部分，在车辆检测区建立跟踪对象，在车辆跟踪区仅仅对跟踪对象进行预测跟踪，从而减少了跟踪的盲目性和算法的计算量，提高算法精度。根据该算法设计的视频检测器已用于实际交通参数的采集，统计的结果表明该算法检测和跟踪的正确率分别高于96%和97%，并具有很好的实时性，能满足管理部门的需求。     

基于场景图划分的无人机影像定位算法

针对大规模长序列无人机（UAV）影像定位中存在的速度慢、误差漂移等问题,结合UAV影像的特点,提出了一种基于场景图划分的UAV影像定位算法。首先,利用全球定位系统（GPS）辅助信息缩小特征匹配的空间搜索范围,从而加速同名点的提取;之后结合视觉一致性和空间一致性来构建场景图,并利用归一化割（Ncut）对其进行划分;接着,对各组场景图进行增量重建;最后,利用光束法平差（BA）融合场景图从而计算出场景的三维模型。此外,在BA阶段,所提算法对代价函数进行扩充,即加入了GPS空间约束信息。在四个UAV影像数据集上的实验结果表明,与COLMAP等多种运动恢复结构（SFM）算法相比,所提算法的定位速度提升了50%,重投影误差减小了41%,定位误差控制在0.5m之内。此外,通过有无GPS辅助下的算法的实验对比,可以得知引入相对和绝对GPS约束的BA有效解决了误差漂移问题,避免了出现歧义性结果,并且极大地减小了定位误差。     

一种波形相似度的车辆检测融合算法

利用磁阻传感器可以实现停车位的车辆{有,无}状态检测,但是磁信号容易受到相邻车位停车干扰而导致误检,针对该问题设计了一种数据融合算法,将车辆磁信号建模为分段线性时间序列,提出了形态距离公式度量时间序列的相似度,计算相邻车位传感器磁信号的波形相似度实现车辆检测,在路面停车区域实地部署了50个磁阻传感器节点,通过长达5个多月的实验,结果表明该算法提高了车辆检测准确率.     

基于Camshift和Kalman滤波的自动跟踪算法

针对传统的Camshift算法在跟踪时需要手动定位目标,在颜色干扰、遮挡等复杂背景中容易跟丢目标的问题,提出了一种基于Camshift和Kalman滤波的自动跟踪算法。首先利用帧间差分法和Canny边缘检测法分割出运动目标的完整区域,然后用提取出的目标区域初始化Camshift算法的初始搜索窗口,从而实现了目标的自动跟踪。当背景中存在相似颜色干扰或者目标被严重遮挡时,采用Kalman滤波与Camshift算法相结合的改进算法进行跟踪。实验结果表明,本文改进算法在目标被严重遮挡、颜色干扰等情况下仍能有效、稳健地跟踪。     

A vehicle detection and tracking method for traffic video based on faster R-CNN

Multimedia Tools and Applications - In this paper we present a vehicle detection and tracking method for traffic video analysis based on deep learning technology. Indeed, with the rapid development...     

融合YOLO检测的多目标跟踪算法

针对目前视频多目标跟踪过程中的遮挡问题,提出了一种融合YOLO v3的多目标检测和跟踪算法,选定基于检测跟踪的框架作为跟踪的整体框架,使用YOLO v3来实现对目标信息的检测工作,在选定某一检测类别的基础上,使用本文提出的跟踪算法,通过数据关联完成对此类别的多目标跟踪,并针对跟踪过程中的目标遮挡问题以及因目标遮挡而引起的轨迹跟踪异常的问题,提出了修正算法。测试视频中被遮挡的大部分目标都能准确地跟踪,但在背景移动时也会发生一部分目标身份互换的情况。所提出的算法在解决多目标跟踪中的遮挡问题时具有一定的准确性和实时性。