利用PLSA技术进行图像分割

引用概率潜语义分析（PLSA）技术作了航空影像图像分割的尝试,介绍了PLSA图像分割的原理方法和分割的具体算法。通过三幅航空影像分割实验,并与生物地理学图像分割（BBOIS）方法和基于吸引子算法的结果作了对比,结果表明,本文方法具有一定的优越性,是一个很有潜力的图像分割方法。     

利用语义的多版本网络覆盖服务模式匹配方法

提出了一种基于节点相似度的语义模式匹配方法,首先对元素标签进行概念化,其次计算元素标签间的语义关系,再次计算节点概念并转化成合取范式CNF,最后利用命题可满足性问题解决器SAT验证节点间的概念关系。实验结果表明,该方法有效提高了匹配的查全率、精度和整体质量。     

MARC: a robust method for multiple-aspect trajectory classification via space,time, and semantic embeddings

ABSTRACT

The increasing popularity of Location-Based Social Networks (LBSNs) and the semantic enrichment of mobility data in several contexts in the last years has led to the generation of large volumes of trajectory data. In contrast to GPS-based trajectories, LBSN and context-aware trajectories are more complex data, having several semantic textual dimensions besides space and time, which may reveal interesting mobility patterns. For instance, people may visit different places or perform different activities depending on the weather conditions. These new semantically rich data, known as multiple-aspect trajectories, pose new challenges in trajectory classification, which is the problem that we address in this paper. Existing methods for trajectory classification cannot deal with the complexity of heterogeneous data dimensions or the sequential aspect that characterizes movement. In this paper we propose MARC, an approach based on attribute embedding and Recurrent Neural Networks (RNNs) for classifying multiple-aspect trajectories, that tackles all trajectory properties: space, time, semantics, and sequence. We highlight that MARC exhibits good performance especially when trajectories are described by several textual/categorical attributes. Experiments performed over four publicly available datasets considering the Trajectory-User Linking (TUL) problem show that MARC outperformed all competitors, with respect to accuracy, precision, recall, and F1-score.     

机载多光谱影像语义分割模型在农田防护林提取中的应用

农田防护林是农田生态系统的屏障,其健康状况的监测与评估在我国北方农田林网管理中尤为重要。本文以新疆生产建设兵团第三师51团为研究区,使用复合翼无人机CW-20搭载Micro MCA12 Snap多光谱相机获取农田防护林的多光谱影像,经辐射校正、裁剪等预处理,通过优选有效特征和模型比较,提出农田防护林提取的有效方法。首先,基于原始12波段,依据相关性系数矩阵和最佳指数因子（Optimum Index Factor,OIF）选取最优3波段和植被指数特征进行组合,构建8种农田防护林提取方案;然后,通过建立语义分割Deeplabv3+模型进行精度评价,得到最优3波段组合6（波长710 nm）、8（波长800 nm）、 11（波长900 nm）波段为最佳特征组合;最后,以最优3波段为基础,将Deeplabv3+模型与U-Net、ENVINet5模型进行对比分析。结果表明：Deeplabv3+模型能够更深层次的挖掘光谱中潜在的信息,相比其他模型,能够较好地处理正负样本不均衡问题,获得最高MIoU值85.54%,比U-Net、ENVINet5的MIoU值则分别高出21.21%、27.19%。该研究结果可为基于多光谱遥感影像的语义分割在农田防护林提取及健康状况监测的应用提供借鉴和参考。     

基于语义规则的高分辨率影像道路交叉口提取方法

为了准确提取遥感影像上道路交叉口目标,提出一种利用特征语义规则从高分辨率影像上提取道路交叉口的方法。该算法构建交叉口模型时将其视为由同质区域像素集合及区域轮廓边界构成的面对象,提取过程分为两步:1)利用辐射、纹理特征语义匹配提取交叉口候选区域;2)通过几何特征语义匹配筛选候选区域、识别交叉口属性。利用多源遥感影像对算法正确性及合理性进行验证,结果表明:算法能准确、完整地提取道路交叉口,可为影像道路网构建提供辅助信息。     

地理空间元数据关联网络的构建

利用资源描述框架（RDF）设计地理空间元数据关联模型,根据地理空间元数据之间的语义关系和语义相关度的计算,以构建以元数据为节点、元数据之间的语义关系为边、语义相关度为权重的关联网络。在这一网络中,一个节点是一个地理空间元数据的资源描述图,包含属性特征（数据来源、空间特征、时间特征、内容）及其关系特征（元数据之间的语义关系、语义相关度）。实验及其分析表明,地理空间元数据关联网络可以有效地支持地理空间数据语义关联检索、推荐等应用,这与传统的基于关键词的元数据检索方式相比,具有更高的准确度。     

面向LOD的地质资料机构知识库语义扩展方法研究

针对地质资料信息服务过程中,存在信息孤岛和数据共享不够等问题,提出了面向开放关联数据LOD的地质资料机构知识库语义扩展方法,并对方法的框架和关键技术进行了研究。首先,基于DSpace构建地质资料机构知识库,自动实现资源描述框架RDF的存储与转化,与LOD形成统一的元数据描述标准。其次,构建地质资料数据的关联模型,明确数据间的语义关系。最后,采用D2RQ平台实现地质资料机构知识库与LOD数据集的语义关联。该方法将进一步加快语义化地质资料信息服务的步伐。     

引入PLSA模型的光学遥感图像舰船检测

提出一种基于概率潜在语义分析(Probabilistic Latent Semantic Analysis,PLSA)的检测算法,首先通过PLSA将目标表述为潜在成分的概率组合,然后利用统计模式识别方法对获取的潜在成分概率进行判别,从而完成最终的检测。其中,生成的潜在成分反映了目标与特征之间相互出现的频率关系,并以潜在成分在目标中概率差异的形式对上述不对应现象给出了直观描述。实验结果表明,所提出的算法对多种复杂情况下的光学图像舰船检测具有很好的准确性和鲁棒性。     

车载激光点云的道路标线提取及语义关联

自动驾驶技术已成为未来智能交通的发展方向之一,高精度地图为L3级及以上自动驾驶实现高精度定位和路径规划提供先验信息,是自动驾驶车辆传感器在遮挡或观测距离受限情况下的重要补充。道路标线的位置和语义信息,比如实线和虚线的绝对位置是高精度地图的基本组成部分。本文从车载激光点云中提取扫描线,根据道路边缘位置几何形态的突变从扫描线中提取道路路面,在此基础上首先利用反距离加权插值的方法把路面点云图像以一定的分辨率转换为栅格图像,其次利用基于积分图的自适应阈值分割方法把栅格图像转化为二值图像,然后利用欧氏聚类的方法从二值图像中提取标线点云,并利用特征属性筛选的方法对提取的标线点云进行语义识别,最后建立交通标线和交通规则之间的语义关联。     

A helium stratified and ingassed lower mantle: resolving the helium paradoxes

It is generally believed a variation of 3He/4He isotopic ratios in the mantle is due to only the decay of U and Th,which produces4 He as well as heat.Here we show that not only3He/4He isotopic ratios but also helium contents can be fractionated by thermal diffusion in the lower mantle.The driving force for that fractionation is the adiabatic or convective temperature gradient,which always produces elemental and isotopic fractionation along temperature gradient by thermal diffusion with higher light/heavy isotopic ratio in the hot end.Our theoretical model and calculations indicate that the lower mantle is helium stratified,caused by thermal diffusion due to*400℃temperature contrast across the lower mantle.The highest3He/4He isotopic ratios and lowest He contents are in the lowermost mantle,which is a consequence of thermaldiffusion fractionation rather than the lower mantle is a primordial and undegassed reservoir.Therefore,oceanicisland basalts derived from the deepest lower mantle with high3He/4He isotopic ratios and less He contents—the long-standing helium paradox,is solved by our model.Because vigorous convection in the upper mantle had resulted in disordered or disorganized thermal-diffusion effects in He,Mid-ocean ridge basalts unaffected by mantle plume have a relatively homogenous and lower!3He/4He isotopic compositions.Our model also predicts that 3He/4He isotopic ratios in the deepest lower mantle of early Earth could be even higher than that of Jupiter,the initial He isotopic ratio in our solar system,because the temperature contrast across the lower mantle in the early Earth is the largest and less4 He had been produced by the decay of U and Th.Moreover,the early helium-stratified lower mantle owned the lowest He contents due to over-degassing caused by the largest temperature contrast.Consequently,succeeding evolution of the lower mantle is a He ingassed process due to secular cooling of the deepest mantle.This explains why significant amount of He produced by the decay of U and Th in the lower mantle were not released,another long-standing heat–helium paradox.