面向地理信息共享的AM/FM/GIS核心语义模型

提出了一种适用于管线行业的AM/FM/GIS数据模型——核心语义模型,对核心语义模型的概念框架,包括定义、编码、空间关系和规则组件环境进行了详细的剖析,得出该数据模型有利于实现管线行业的信息共享、提高数据录入精度、完善地理要素的表达,知识的可扩展性和规则组件的可重用性保证了核心语义模型具有更开放灵活的用户参与机制,最后提出了下一步研究的方向。     

Deep Fractionation of Clinopyroxene in the East Pacific Rise 13°N: Evidence from High MgO MORB and Melt Inclusions

Mid-ocean ridge basalts (MORBs) from East Pacific Rise (EPR) 13°N are analysed for major and trace elements, both of which show a continuous evolving trend. Positive MgO–Al2O3 and negative MgO–Sc relationships manifest the cotectic crystallization of plagioclase and olivine, which exist with the presence of plagioclase and olivine phenocrysts and the absence of clinopyroxene phenocrysts. However, the fractionation of clinopyroxene is proven by the positive correlation of MgO and CaO. Thus, MORB samples are believed to show a “clinopyroxene paradox”. The highest magnesium-bearing MORB sample E13-3B (MgO=9.52%) is modelled for isobaric crystallization with COMAGMAT at different pressures. Observed CaO/Al2O3 ratios can be derived from E13-3B only by fractional crystallization at pressure >4 ±1 kbar, which necessitates clinopyroxene crystallization and is not consistent with cotectic crystallization of olivine plus plagioclase in the magma chamber (at pressure ~1 kbar). The initial compositions of the melt inclusions, which could represent potential parental magmas, are reconstructed by correcting for post-entrapment crystallization (PEC). The simulated crystallization of initial melt inclusions also produce observed CaO/Al2O3 ratios only at >4±1 kbar, in which clinopyroxene takes part in crystallization. It is suggested that MORB magmas have experienced clinopyroxene fractionation in the lower crust, in and below the Moho transition zone. The MORB magmas have experienced transition from clinopyroxene+plagioclase+olivine crystallization at >4±1 kbar to mainly olivine+plagioclase crystallization at <1 kbar, which contributes to the explanation of the “clinopyroxene paradox”.     

A statistical adjustment of Haldorsen's conditioned Boolean simulation algorithm

This paper presents a new simulation algorithm for generating realizations of a Boolean model for sandstone reservoirs conditional on sandstone body intersections in wells. It is a statistically corrected version of the conditional simulation algorithm originally proposed by Haldorsen. In previous work it was shown that the conventional algorithm does not reproduce the correct statistics for sandstone body size at the well locations. The simulation of a Boolean model, given grain intersections on line transects, must be in accordance with the conditional distribution of the model, which implies that sandstone bodies intersected by wells and sandstone bodies in the interwell area should be simulated independently. Based on the conditional distribution a simulation algorithm is developed, which is compared to Haldorsen's algorithm by simulating an outcrop section of fluvial sandstone deposits. Simulations are conditioned on data of two fictitious wells. It turns out that the adjusted algorithm gives better results for the sand fraction that is connected to a set of wells, and also for the sand fraction that would be connected to an infill well.     

深度融合网结合条件随机场的遥感图像语义分割

为了充分利用遥感图像中丰富的细节信息和上下文信息,提高图像语义分割精度,提出一种深度融合网结合条件随机场模型的遥感图像语义分割方法。方法在全卷积神经网络框架中增加反卷积融合结构,搭建深度融合DFN (Deep Fusion Networks)网络,通过深层网络自动获取多尺度特征,避免人工设计和选择特征,提高模型的泛化能力;同时借助反卷积融合结构,利用多尺度信息,将浅层细节信息和深层语义信息相融合,提高模型的处理精度。由全连接条件随机场引入空间上下文信息,更好地定位边界,得到最终的语义分割结果。在遥感图像数据集上的实验结果显示:(1)随着不同尺度细节信息的融入,结果的边缘轮廓越精确、接近标签图像;(2)增加了空间上下文信息后,语义分割结果边缘更细化、准确,精度更高。实验表明,该方法可以有效提高遥感图像语义分割的精度,改善结果的过平滑现象。     

深度卷积神经网络支持下的遥感影像语义分割

针对高分遥感影像语义分割面临的类别不平衡和上下文信息利用不充分问题，本文提出了一种优化的DeeplabV3+算法。首先通过修改交叉熵损失函数，解决数据不平衡问题；其次使用Vortex Pooling取代ASPP模块提高上下文信息；然后采用多尺度输入充分利用图像的多尺度信息，并用投票策略进行特征融合提高图像分割准确性；最后使用形态学作后处理消除拼接痕迹和噪声。在CCF大赛的数据集上进行训练，并与其他经典语义分割算法进行比较。试验结果表明，该算法充分利用上下文信息，有效减少了错误分类，且使分割边界更精确，尤其对于线状目标的捕捉能力更强；在整幅测试影像上的MIoU可达85.21%，明显优于SegNet、U-Net算法。     

交通监控视频图像语义分割及其拼接方法

视频拼接是图像拼接的外延,在场景监控、目标识别等应用中发挥着重要作用。传统视频拼接算法多要求视频间具有较大重叠区域且特征点匹配过程中只顾及图像几何特征,当处理交通监控视频时,会因不同摄像头之间重叠区域极小或主光轴之间夹角较大而导致无法拼接或图像变形较大。为此,本文提出一种交通监控视频图像语义分割及其拼接方法。首先,利用边缘角度二阶差分直方图算法自动识别多视频交汇区域的正射影像,并将其作为拼接背景图像;然后,基于全卷积神经网络对正射影像和视频图像进行语义分割,提取图像中的交通专题语义;最后,以交通专题语义作为约束进行特征点匹配,将各个交通监控视频匹配至背景正射影像,实现监控区域视频拼接。采用山东省某市实际视频数据进行试验验证,结果表明对于重叠区域较小的监控视频,本文方法可获得较好地拼接图像,同时可有效提高特征点匹配的准确度。     

The diverse and contested meanings of sustainable development

We provide a heuristic framework that can be used as a lens for understanding the arguments being presented in the papers which comprise this special issue on sustainable development and environmental issues in Great Britain. This framework can also be used as an introduction to the wider literature on sustainable development because it is designed to bring greater clarity to this large, diverse and rapidly expanding field of enquiry populated by heterogeneous discourses, multiple approaches and a variety of recommendations as to the ways forward.     

空间信息语义网格

在分析本体和网格技术的基础上,针对空间信息系统对信息语义的特殊要求,提出了空间信息语义网格的概念,并用本体构造了空间信息语义网格体系结构。     

FME在空间数据加工项目中的应用

针对福建电信数据加工项目,介绍了如何利用基于语义的数据转换处理软件FME,实现各种繁杂空间数据处理,包括开发数据格式插件、集成的数据批处理程序,为空间数据加工项目提供了一个很好的思路。     

基于本体(Ontology)的空间信息互操作与集成方法研究

随着Semantic Web技术的发展,基于Ontology的空间信息服务是分布式网络环境下异构空间信息资源共享与互操作实现的可行途径.在对空间信息共享与互操作发展历程、Ontology的基本理论及国内外基于Ontology的空间信息共享与互操作研究现状进行系统全面总结概括的基础上,提出了Ontology驱动的空间信息共享与互操作系统的体系框架、服务流程,并提出了今后研究的重点.基于该服务框架构建的信息系统能将分散在Internet上的各种空间资源信息动态地集成在一起,实现分布式环境下异构空间信息资源的共享与互操作.