度量空间上的位势空间容量与Hausdorff测度

讨论了度量空间上的位势空间,并利用容量的概念建立其与Hausdorff测度之间的联系.     

空时级联滤波红外点目标检测

时域递归最大值滤波是一种较易实现的红外小目标检测方法,但是在递归过程中存在目标膨胀的缺陷,影响了该方法的应用。通过研究分析,提出了条件最小值滤波替换最小值滤波抑制目标膨胀,较好解决了目标膨胀问题。结合空域最大中值滤波预测背景,将点目标和强噪声保留在预测残差中,再通过递归最大值滤波对预测残差进行时域递归处理,以完成能量累积提高信噪比,设计完成的算法实现了对尺度为 1 个像素,运动速度小于 1 像素每帧的点目标的可靠检测。     

单台光电经纬仪对目标稳定跟踪的研究

针对单台光电经纬仪目标提取特性短暂失去的情况下设备平稳过渡的问题,详细研究了目标在单台经纬仪中的运动轨迹。利用已有的数据对目标的运动轨迹进行分段最佳多项式逼近与外推,实现设备的平稳过渡。结果表明,此方法可以比较理想地解决目标短时间(2-3s)内丢失情况下的对目标的稳定跟踪问题。     

Autonomous target tracking and following mobile robot

Tracking and following a moving target in real time is a very challenging task in autonomous mobile robot applications, due to the unstructured and unknown environment. In this article, a real time, autonomous, dynamic, whole-field target tracking system (360° coverage) is developed based on a pan/tilt/zoom CCD vision system. The vision system scans and locks the pose of the moving target and commands the tracking mobile robot to follow the target while avoiding obstacles. Bayes’ theorem-based foreground–background segmentation algorithm is applied for motion detection. The color-based particle filter algorithm is used to track the moving object. The system steers the tracking mobile robot toward the moving target based on the angular difference obtained from the tracking module. Fuzzy logic is applied and a sensor data fusion algorithm is developed to reduce the ultrasonic phantom effect of obstacle detection using eight ultrasonic sensors. Multiple behaviors are integrated through the proposed system. Experimental results show the proposed system can successfully track and follow a moving target and avoid obstacles in real time. The tracking accuracy is higher than 80%.     

弱小目标检测与跟踪算法

低对比度小目标检测与跟踪算法的研究是电视跟踪领域的关键技术之一。针对弱小目标的目标特性，按USAN原则进行目标检测，利用目标的特征参数及目标运动的一致性、连续性排除噪声干扰，实现对目标的稳定跟踪。实验采用VC＋＋仿真平台验证该算法的可行性和有效性，并移植到DSP专用图像处理平台上，达到了工程的5可靠性与实时性要求。     

Dynamic target tracking with fringe-adjusted joint transform correlation and template matching

Target tracking in forward-looking infrared (FLIR) video sequences is a challenging problem because of various limitations such as low signal-to-noise ratio (SNR), image blurring, partial occlusion, and low texture information, which often leads to missing targets or tracking nontarget objects. To alleviate these problems, we developed a novel algorithm that involves local-deviation-based image preprocessing as well as fringe-adjusted joint-transform-correlation--(FJTC) and template-matching--(TM) based target detection and tracking. The local-deviation-based preprocessing technique is used to suppress smooth texture such as background and to enhance target edge information. However, for complex situations such as the target blending with background, partial occlusion of the target, or proximity of the target to other similar nontarget objects, FJTC may produce a false alarm. For such cases, the TM-based detection technique is used to compensate FJTC breaking points by use of cross-correlation coefficients. Finally, a robust tracking algorithm is developed by use of both FJTC and TM techniques, which is called FJTC-TM technique. The performance of the proposed FJTC-TM algorithm is tested with real-life FLIR image sequences.     

A similarity coefficient measure and machine-parts grouping in cellular manufacturing systems

For more than three decades, similarity coefficient measures one of the important tools for solving group technology problems have gained the attention of the research community in cellular manufacturing systems. A new similarity coefficient measure that uses a set of important characteristic properties for grouping is developed here for use as an intermediate tool to form cohesive cells. A mathematical model that uses this similarity coefficient for optimally solving the cell-formation problems in cellular manufacturing is developed. A heuristic procedure that improves the optimal methodology in term of solution capability of the large instances is devised for an efficient solution. Both the optimal methodology and the heuristic are applied to some well-known problems from literature to compare the grouping efficiencies. The similarity coefficient and the solution methodologies developed are able to solve the cell formation problems efficiently.     

复杂场景下的红外目标检测

主流的目标检测网络在高质量RGB图像上的目标检测能力突出,但应用于分辨率低的红外图像上时目标检测性能则有比较明显的下降。为了提高复杂场景下的红外目标检测识别能力,本文采用了以下措施：第一、借鉴领域自适应的方法,采用合适的红外图像预处理手段,使红外图像更接近RGB图像,从而可以应用主流的目标检测网络进一步提高检测精度。第二、采用单阶段目标检测网络YOLOv3作为基础网络,并用GIOU损失函数代替原有的MSE损失函数。经实验验证,该算法在公开红外数据集FLIR上检测的准确率提升明显。第三、针对FLIR数据集存在的目标尺寸跨度大的问题,借鉴空间金字塔思想,加入SPP模块,丰富特征图的表达能力,扩大特征图的感受野。实验表明,所采用的方法可以进一步提高目标检测的精度。     

免疫粒子滤波在声呐图像目标跟踪中的应用

基于前视声呐图像序列,研究并实现了经免疫算法优化的粒子滤波水下目标跟踪。声呐图像分割成二值图后,提取目标的区域形状特征以构建观测模型,设计目标模板自适应更新方法;将免疫算法的克隆与变异思想引入到粒子滤波中以解决粒子退化问题。对两组水下运动物体的跟踪实验表明,即使目标存在一定形变与干扰,文中的免疫粒子滤波算法仍能以较高的精度跟踪到目标真实运动轨迹;相比于传统粒子滤波算法,稳定性也更强。     

Infrared target detection with probability density functions of wavelet transform subbands

We report the development of a wavelet multiresolution texture-based algorithm that uses the probability density functions (PDFs) of the subband of the wavelet decomposition of an image. The moments of these pdfs are used in a clustering algorithm to segment the targets from their background clutter. Using the tools of experimental methodology, we evaluate the performance of this algorithm on real infrared imagery under varying algorithm parameter sets as well as scene, image, and false-alarm conditions. We estimate a set of multidimensional predictive analytic performance models that relate the detection probabilities as functions of false alarm, algorithm internal parameter, target pixel number, target-to-background interference ratio, target-interference ratio, and Fechner-Weber and local entropy metrics in the scene. These models can be used to predict performance in regions were no data are available and to optimize performance by selection of the optimum parameter and constant false-alarm values in regions with known scene and metric conditions.     

智能目标检测与跟踪系统的设计与实现

为了实现复杂背景下对任意目标的智能检测与稳定跟踪,设计了基于双DSP结构的集视频处理、云台运动控制于一体的实时电视跟踪系统.系统采用模块化硬件结构,并提供直接用户接口可使系统脱离上位机独立工作.同时改进了目标识别算法,提出一种基于区域分割的相关模板提取方法.利用自适应空域滤波器分割出目标区域,再用二次投影法快速提取目标,并以模板评价函数为准则实时更新模板.具有模板自适应性强,目标判断准确,跟踪鲁棒性好等优点.外场实验表明,在目标变化和有物体遮挡干扰的情况下,系统能准确提取任意指定目标,并保持连续稳定的跟踪.     

EKF和PF算法及其在目标跟踪中的应用

基于贝叶斯滤波的目标跟踪原理,介绍了扩展卡尔曼滤波(Extended Kalman Filter,EKF)和粒子滤波(ParticleFilter,PF)的基本思想和算法实现步骤。在非线性环境下对比分析了EKF算法和PF算法的估计精度,并给出两种方法的适用条件。EKF算法采用Taylor展开的线性变换来近似非线性模型,而PF算法采用一些带有权值的随机样本来表示所需要的后验概率密度。仿真结果表明,在强非线性非高斯环境下,PF算法的跟踪性能远优于EKF算法,当系统非线性强度不大时,EKF算法和PF算法的估计精度相差不大,但PF算法计算复杂,跟踪时间长,实时性差。     

基于双曲调频距离误差的目标跟踪修正算法

双曲调频信号的多普勒不变性可使运动目标匹配滤波输出的能量损失达到最小，在运动目标检测与参数估计中具有独特优势，已广泛应用于新型主动声呐。然而其多普勒不变性在提高检测能力的同时，会引入目标距离估计误差，从而影响主动声呐跟踪性能。文章在分析双曲调频信号距离估计误差的基础上，提出了目标跟踪算法的误差修正方法，通过将距离估计误差公式与传统观测方程及滤波算法进行融合，实现目标跟踪算法的误差补偿，显著提高了算法的跟踪性能，尤其是对快速运动目标的跟踪性能。     

A universal measure of the conformity of a mesh with respect to an anisotropic metric field

In this paper, a method is presented to measure the non‐conformity of a mesh with respect to a size specification map given in the form of a Riemannian metric. The measure evaluates the difference between the metric tensor of a simplex of the mesh and the metric tensor specified on the size specification map. This measure is universal because it is a unique, dimensionless number which characterizes either a single simplex of a mesh or a whole mesh, both in size and in shape, be it isotropic or anisotropic, coarse or fine, in a small or a big domain, in two or three dimensions. This measure is important because it can compare any two meshes in order to determine unequivocally which of them is better. Analytical and numerical examples illustrate the behaviour of this measure. Copyright © 2004 John Wiley & Sons, Ltd.     

Gabor小波目标特征提取和跟踪方法的研究

为了解决目标在发生旋转、尺度及平移等几何变形时的自动跟踪问题,提出 Sobel 边缘检测与 L-M 优化结合的 Gabor 小波目标提取与跟踪方法。该方法自动寻找目标的小波特征点,在跟踪过程中不断优化特征模板的旋转、尺度以及平移等几何变形参数,使得小波特征向量值在最小平方和意义上与初始值相匹配,实现目标跟踪。实验结果表明,该算法能够有效地提取目标特征,在目标尺度变化达到 50%,角度变化 10o,旋转 90o等不利条件下实现了可靠跟踪。     

基于SSD-CF的无人艇目标检测跟踪方法

针对目前目标检测和目标跟踪算法对无人艇运算配置要求高、速度慢等问题,该文一种提出基于SSD-CF的无人艇目标检测跟踪方法。利用MobileNets结构结合SSD目标检测算法构建轻量级卷积神经网络,实现无人艇的水面目标检测。目标检测结果作为相关滤波CF目标跟踪算法的初始输入,并在目标跟踪过程的保障其有效性。通过MODD水面船只视频数据实验表明,SSD-CF方法融合目标检测与目标跟踪算法,可有效地降低对运算力的要求,提升目标检测跟踪速度和目标位置的稳定连续性。     

Error-space estimation method and simplified algorithm for space target tracking

To improve space target tracking precision and the stability of mobile optoelectronic tracking equipment, an error-space estimation method based on the Kalman filter is discussed, and a simplified algorithm is presented to reduce calculation cost. Based on an available measurement of a space target without sufficient validity and accuracy, the actual position related to the tracking equipment is decomposed to an earlier offline prediction of the kinetic model method and prediction errors. By regarding prediction errors as the motion of a weak maneuver target, the errors can be estimated more accurately in error space. By synthesizing estimation of the errors and offline prediction, the space target position is obtained with higher accuracy to improve tracking performance.     

复杂背景下飞行目标的快速提取与跟踪

为了对飞行目标(典型目标为导弹和飞机)进行跟踪,利用差分图像结合最大空穴检出算法进行目标的提取和捕获,以模板评价函数为准则对模板进行适时更新,利用归一化的互相关系数确定目标在图像中的位置.为了提高速度,尽量采用IPP函数处理大部分的算法,相比较于传统的函数,IPP函数大幅度地提高了运算速度.实验证明,该方法能够在目标变化较大、图像模糊、对比度较低的情况下准确识别并跟踪目标,处理速度基本满足了使用要求.     

Toward establishing a unified metric for performance and learning goal orientations

Findings in the goal orientation literature are convoluted due to sample and scale dependent interpretations and potentially the use of inappropriate measurement models for ordinal data. The purpose of this investigation was to explore the possibility of developing a common metric in order to enhance communication among goal orientation researchers using different scales. A simultaneous calibration of four commonly used goal orientation scales was undertaken to investigate if item responses meet the unidimensionality requirement of Rasch models and hence provided a common metric. Participants were 305 third year psychology students at a large University in France. Results indicated that 30 of 36 learning goal and 28 of 33 performance goals items from the four scales defined unidimensional constructs. Correlations between person measures from the four scales and the simultaneous calibration for both learning goal and performance goals support the interpretation of unidimensional variables. Raw score to interval measure conversion tables provide goal orientation researchers using different scales a common metric for interpretation and statistical analyses. Limitations and extensions of this research are discussed.