A Coupled Visual Representation based on Local Patch and Context for Single Target Tracking

来源 :第20届中国系统仿真技术及其应用学术年会(20th CCSSTA 2019) | 被引量 : 0次 | 上传用户:chloexg
下载到本地 , 更方便阅读
声明 : 本文档内容版权归属内容提供方 , 如果您对本文有版权争议 , 可与客服联系进行内容授权或下架
论文部分内容阅读
  Visual representation has an important influence on tracking performance.Unfortunately,most of the existing methods construct the visual representation only focus on the global target information, which ignored the information of the internal structure and the background of the target.To improve the accuracy of visual representation, a novel approach via local parts and context jointly is proposed in this paper.First, the target is decomposed into multiple independent patches based on the spatial structure, and each patch learns the filtering response separately.Second, the contextual information is introduced into the representation framework to migrate the background influences.With the introduction of collaborative representation strategy, the impact of background noise can be effectively suppressed.By conduct extensive experiments on several challenging benchmark datasets, the proposed approach demonstrates comparable performance against several state-of-the-art methods.
其他文献
For some rodent mammals when they foraging or looking for a target, the positions and headings in their brain cells are distributed as a hexagonal raster shape.Hex-grid maps used in the navigation of
Nowadays due to the advancement of technology, especially in the field of computer science,quality of the maps, resolution as well as the image details, has been remarkable.Although having a more deta
随着深度学习的快速发展,基于深度神经网络的目标检测模型被不断提出并且检测的效率和精度均得到大幅提升.然而,目前的环境目标检测方法在复杂场景下的性能仍然受限.多传感器信息融合被认为是进一步实现目标检测技术突破的重要途径.本文针对环境感知中的车辆检测技术,着重介绍了融合视觉信息和激光信息的多视图目标检测(AVOD)模型的框架,对其工作原理进行说明,探究了决定网络模型性能的关键因素.最后,论文在KITT
Simultaneous localization and mapping (SLAM)has made astonishing progress over the last 30 years and ORB-SLAM system is definitely one of the most essential and popular frameworks so far.The ORB-SLAM
本文描述飞机飞行控制系统电缆设计使用的工具、电线材料选取、线束长度确定,以及基本的画法,供飞机飞行控制系统及其他系统电缆设计时参考.
本文针对智能集群区域探测问题,研究了一种基于质心Voronoi图法的多智能体区域覆盖控制策略.针对质心Voronoi图法的收敛性与初始构型耦合这一问题,采用先分散、后聚拢两步法运动,使得多智能体分布效果明显提升.定义了区域覆盖问题的目标代价函数,以代价函数的函数最优为目标,进行收敛性仿真研究.仿真结果证明了控制策略的可行性和有效性.
Edge is a significant feature for many computer vision applications.Existing edge detectors obtain a dense edge map depended on the contrast and gradient cues in an image.However, the dense edge map n
网络战、网络恐怖主义和网络犯罪是对国家安全的严重威胁.这种威胁使得对熟练网络安全人员的需求日益增加,而现有的培训方式却无法训练出足够的专业的网络安全人员来满足这一需求.因此,迫切需要迅速建立创新、高效的网络安全教育和培训方案.美国开始使用网络安全员学院课程(COAC)进行网络安全员培训来解决这一问题,但只能满足部分的网络安全教育和培训需求.还应该考虑其他方法,如设计良好的游戏以及游戏化的学习环境.
随着卫星遥感图像分辨率的提升,感兴趣目标在图像中所占尺寸越来越小,从地面复杂背景中快速提取出感兴趣的目标依然具有挑战性.本文提出了一种基于局部学习方法的直推支持向量机的遥感图像目标提取方法.该方法将半监督学习方法和局部学习方法相结合,能够有效利用已知目标样本的直推支持向量机,实现对小样本遥感目标识别的分类器构建.实验结果表明,该方法在高分辨率卫星遥感图像上的海上目标提取上比其他方法有更好的性能.
针对单部火控雷达跟踪目标时跟踪范围的局限性,以两部火控雷达为例提出了一种多雷达接力式跟踪方法并建立了跟踪模型,加快了下一接力雷达航迹滤波的收敛速度.通过对比分析接力跟踪时雷达间的不同分布对跟踪航迹质量的影响,得到最优的接力跟踪距离分布.同时此模型也适用于光电跟踪等其他多传感器目标跟踪系统,本研究为多传感器接力跟踪的距离分布优化提供了方法支撑.