2008年以来地图学眼动与视觉认知研究新进展

地图是人类日常生活中不可或缺的工具。地图认知作为研究人对地图的感知、学习、记忆、推理和决策的科学,一直以来是地图学基础理论的重要内容。作为地图认知研究的核心,近年来地图视觉认知在相关学科（如心理学、认知科学和计算机视觉）和新的研究手段（如眼动跟踪、脑电和核磁）的促进下取得了一系列新的研究成果。尤其是眼动跟踪方法作为地图视觉认知研究的重要手段之一,越来越受到研究者的关注,为地图视觉认知研究的定量化、实用化提供了有力支撑。本文系统地梳理了2008年以来眼动跟踪方法在地图视觉认知研究中取得的新进展,并将其归纳为6个发展趋势,分别包括：① 刺激材料：从静态地图到动态交互地图;② 研究范围：从地图认知到地图空间认知;③ 实验环境：从实验室环境到真实环境;④ 地图维度：从二维地图到三维地图;⑤ 个体差异：从单一维度到多维度;⑥ 研究目的：从规律探究到实践应用。本文最后总结了未来研究面临的挑战、难点和可能的解决方法,期望能在地图视觉认知研究中起到抛砖引玉的作用。     

Particle tracking experiments on a model of the Okhotsk Sea: Toward oil spill simulation

Particle tracking experiments were conducted for the Sea of Okhotsk using a three-dimensional ocean circulation model, as a step toward the simulation of oil spills. The model’s reproducibility is first examined in detail. Comparison with surface drifter and moored ADCP data shows that the model successfully reproduces the velocity field over the shelves, particularly in the weak stratification period. This is because the current variability is simply determined by integration of the alongshore component of the wind stress over the coast from which arrested topographic waves propagate. Good agreement even in the ice-covered period implies that the neglect of sea ice in the model is not a problem for reproduction of the current over the shelves. Good agreement also supports the correction of ECMWF wind speed by a factor of 1.25. A series of particle tracking experiments was carried out to examine the case of particles released from the Sakhalin oil field at depths of 0 m and 15 m. Regardless of the deployment month and year, most particles at depth 15 m are transported southward along the Sakhalin coast, in accordance with the abrupt intensification of the East Sakhalin Current in October, finally arriving offshore of Hokkaido in November–January. Particles at the surface, which are affected by wind drift in addition to the ocean current, show larger yearly variability. In years when the offshoreward-wind dominates, the particles would be advected out of the mainstream of the current and would not be transported offshore of Hokkaido.     

RAPTOR‐UAV: Real‐time particle tracking in rivers using an unmanned aerial vehicle

River system measurement and mapping using UAVs is both lean and agile, with the added advantage of increased safety for the surveying crew. A common parameter of fluvial geomorphological studies is the flow velocity, which is a major driver of sediment behavior. Advances in fluid mechanics now include metrics describing the presence and interaction of coherent structures within a flow field and along its boundaries. These metrics have proven to be useful in studying the complex turbulent flows but require time‐resolved flow field data, which is normally unavailable in geomorphological studies. Contactless UAV‐based velocity measurement provides a new source of velocity field data for measurements of extreme hydrological events at a safe distance, and could allow for measurements of inaccessible areas. Recent works have successfully applied large‐scale particle image velocimetry (LSPIV) using UAVs in rivers, focusing predominantly on surficial flow estimation by tracking intensity differences between georeferenced images. The objective of this work is to introduce a methodology for UAV based real‐time particle tracking in rivers (RAPTOR) in a case study along a short test reach of the Brigach River in the German Black Forest. This methodology allows for large‐scale particle tracking velocimetry (LSPTV) using a combination of floating, infrared light‐emitting particles and a programmable embedded color vision sensor in order to simultaneously detect and track the positions of objects. The main advantage of this approach is its ability to rapidly collect and process the position data, which can be done in real time. The disadvantages are that the method requires the use of specialized light‐emitting particles, which in some cases cannot be retrieved from the investigation area, and that the method returns velocity data in unscaled units of px/s. This work introduces the RAPTOR system with its hardware, data processing workflow, and provides an example of unscaled velocity field estimation using the proposed method. First experiences with the method show that the tracking rate of 50 Hz allows for position estimation with sub‐pixel accuracy, even considering UAV self‐motion. A comparison of the unscaled tracks after Savitzky–Golay filtering shows that although the time‐averaged velocities remain virtually the same, the filter reduces the standard deviation by more than 40% and the maxima by 20%. Copyright © 2017 John Wiley & Sons, Ltd.     

移动增强现实跟踪注册技术概述

随着移动设备软硬件性能的不断提高,将增强现实技术与移动终端相结合已成为国内外研究的热点问题.三维跟踪注册技术是移动增强现实中的关键技术之一.文中综述了移动增强现实中的三类跟踪注册方法：基于传感器、基于机器视觉和基于混合的跟踪注册,并对各方法的特点进行了详细的探讨和比较.最后对移动增强现实跟踪注册技术进行了总结与展望.     

郑州CINRAD风暴追踪产品误差统计分析

利用2009—2016年5—9月郑州新一代天气雷达资料,对5615个单体回波按不同预报时效、不同生命史和不同类型强对流的风暴追踪信息产品进行统计,较系统地分析了单体回波的距离误差和方向误差,探讨了实况与预报的偏差方向。结果表明:30和60 min距离误差分别为9. 3和16. 9 km,方向误差变化不大,随着预报时效的延长,距离误差增大;生命史≤1h、1～2 h和超过2 h的30 min预报平均距离误差分别为9. 8、8. 8和7.7 km,方向误差分别为25. 2°、25. 7°和22.8°;生命史在1～2 h和超过2 h的60 min预报平均距离误差分别为17. 3和15. 9 km,方向误差为24. 4°和22. 7°,随着单体生命史的延长,方向和距离误差呈减小趋势;西北气流型、西南气流型两种类型单体回波和整体误差一致,两种类型回波单体30和60 min方向和距离误差基本相当;不同类型、不同预报时效统计均为实况偏于预报路径右侧的次数多于偏左的次数,30和60 min偏右比例分别为57. 5%和55.6%。     

基于OLDM与贝叶斯估计的鲁棒视觉跟踪

提出了一种在线学习判别式模型OLDM(online learning discriminative model),并结合贝叶斯估计实现了对视觉运动目标的鲁棒跟踪。首先,通过对初始化的跟踪区域进行样本标记与聚类分析得到目标的判别式模型;然后,利用该模型在预测的跟踪区域内计算目标的似然分布;最后,在贝叶斯框架下完成目标状态的确定并对模型进行学习与更新。算法通过在线学习适时更新目标模型,增强了算法对目标表观变化的适应性。实验结果表明,本文算法能够有效地适应目标表观特征的复杂变化,对目标的尺度、光照、遮挡以及非刚性形变等具有较强的鲁棒性,算法的跟踪精度与稳定性比当前主流算法均有一定提高。     

基于视觉图像的震区人员搜救方法研究

传统基于信号传感技术的震区人员搜救方法,利用惯性导航传感技术融合移动距离和方向角对待搜救人员进行定位,容易受到天气状况不理想以及局部遮挡的影响,其搜救效率和精度低。提出基于视觉图像的震区人员搜救方法,利用图像采集设备采集震区初始视觉图像,采用小波降噪法降噪处理,提升可辨程度。通过色彩对比方法提取降噪后震区人员视觉图像的特征,并与震区原始图像特征对比,获取震区待搜救人员候选图像;根据候选图像分别采用Kalman滤波跟踪算法和Mean shift跟踪算法,在天气状况不理想以及局部遮挡的复杂震区环境中跟踪搜救人员目标。实验结果表明,所提方法查全率保持在98.5%以上,平均准确率约为98%,人员搜救平均时间约为23 s,说明所提方法能够进行高效、准确的震区人员搜救。     

GOCE卫星重力计划及其应用

基于CHAMP和GRACE卫星，GOCE(Gravity Field and Stead—state Ocean Circulation Explore)是欧空局(ESA)的一颗重力场和静态洋流探测卫星。利用它可得到空间分辨率为200—80km的全球重力场模型和1cm精度的大地水准面．简要介绍了目前重力卫星的发展现状与其局限性，详细叙述了GOCE卫星的组成、科学目标、测量原理、在地球物理等学科中的重要应用，并提出GOCE等重力卫星资料在我国的应用设想。     

多遥感卫星接收系统和其数据接收

该文叙述了单遥感卫星接收系统升级改造为多卫星接收系统所进行的研制工作。包括天线跟踪系统与高频信道的改造，卫星捕获软件的开发，中频接收解调设备的研制以及系统的总装调试等。     

基于PS-VF-TBD的地波雷达船只目标航迹起始方法

针对地波雷达低信噪比目标信号"闪烁"导致的航迹起始困难问题,本文提出了一种基于PS-VF-TBD的航迹起始方法.该方法综合利用了地波雷达回波谱中目标信号的展宽特征和运动特性,利用PS-VF-TBD方法对运动目标能量有效积累的特点,提高对疑似目标多帧信号间的关联概率,进而提高对真实目标的搜索概率和航迹起始成功率.通过地波...