基于压缩感知的Curvelet域联合迭代地震数据重建

由于野外采集环境的限制,常常无法采集得到完整规则的野外地震数据,为了后续地震处理、解释工作的顺利进行,地震数据重建工作被广泛的研究.自压缩感知理论的提出,相继出现了基于该理论的多种迭代阈值方法,如CRSI方法（Curvelet Recovery by Sparsity-promoting Inversion method）、Bregman迭代阈值算法（the linearized Bregman method）等.CSRI方法利用地震波形在Curvelet的稀疏特性,通过一种基于最速下降的迭代算法在Curvelet变换域恢复出高信噪比地震数据,该迭代算法稳定,收敛,但其收敛速度慢.Bregman迭代阈值法与CRSI最大区别在于每次迭代时把上一次恢复结果中的阈值前所有能量都保留到本次恢复结果中,从而加快了收敛速度,但随着迭代的进行重构数据中噪声干扰越来越严重,导致最终恢复出的数据信噪比低.综合两种经典方法的优缺点,本文构造了一种新的联合迭代算法框架,在每次迭代中将CRSI和Bregman的恢复量加权并同时加回本次迭代结果中,从而加快了迭代初期的收敛速度,又避免了迭代后期噪声干扰的影响.合成数据和实际数据试算结果表明,我们提出的新方法不仅迭代快速收敛稳定,且能得到高信噪比的重建结果.     

对文化微空间的感知与认同研究——以广州太古汇方所文化书店为例

自20 世纪70 年代以来,经济文化全球化成为微空间生产与重构的重要动力。而随着生产方式和生活方式的改变,以符号和审美消费为主的后现代消费社会范式的崛起,促使文化成为行就消费空间的主要手段之一。广州方所文化书店就是在这样的背景下催生的新生事物。贴上文化标签后的方所文化书店,一反实体书店在现代消费社会中逐步被边缘化的窘境,出现伊始就受到各界的广泛关注。研究剖析了方所文化书店这类实体书店文化身份的展演（performance）,并对消费者的感知与认同进行了解读。结果显示,通过经营者对文化手段的操纵,方所文化书店被形塑成一个世界文化的微观缩影,同时也成为一个具有尊贵与典雅特质的文化新空间。而消费者通过对实体微空间的环境、商品文化及人群品质的感知,也认同了这种新的文化消费空间。实体书店在标签化的文化表征与文化变迁中完成自我救赎,同时也被消费者重新认同。     

基于XSLT的图示表达规则的设计与实现

地理信息的符号化在ISL/dIS19117中称为地理信息的图示表达,包括表达目录、表达规范和表达服务3个主要部分。其中表达目录组织一系列的表达规则,用于测试地理要素的空间要素与地图显示环境．如显示时间、显示比例尺、地图显示的方向等,并调用表达服务。表达规范描述地图符号的图形参数,为表达服务提供参数属性值。表达服务提供符号生成操作,其函数参数值来源于表达规范。表达规范与表达目录相对独立,共同构成了表达信息。同一表达目录可以调用不同的表达规范,从而生成不同的地图。同样表达规范也可以被不同的表达目录调用。表达信息与地理要素独立,并分开存储在不同的地方。图示表达规则是根据制图知识,推理地理信息的几何与属性,并从图式规范中匹配出相应的地图符号。本文结合表达规则与可扩展样式单语言转换(XSLT．eXtensible Style Sheet Language Transformation)的特点,提出了基于XSLT的图示表达规则模型,采用XSLT描述表达规则,能便于表达规则的网络共享与实时交互。并以地理标记语言GML为对象．通过XSLT将GML转换为基于XML的自定义地图制图标记语言MapML,在此基础上,对MapML进行了符号化,生成SVG地图。实验表明通过XSLT能有效测试与分析地理数据的空间特性与专题属性,并顾及了地图显示比例尺。     

基于人工神经元网络的空间知识表达与推理方法研究

分析了智能决策支持系统的结构,通过比较指出了基于符号的知识表达与推理方法的不足,阐述了基于神经网络的知识表达与推理的方法,并认为神经网络方法可作为智能空间决策支持系统中符号方法的替代或补充,并通过实例验证了基于神经网络的不确定空间知识表达与推理的有效性.     

Dynamic responses under the excitation of pulse sequences

This paper studies the dynamic responses of SDOF system under pulse-dominant excitations. The purpose of the study is to prepare for scrutiny of some near-field pulse-dominant ground motions and their potential to cause structural damage. Extending the single pulse dynamics, we consider the effect of pulse sequences. This kind of excitation was particularly obvious in some of previous earthquakes such as Northridge (1994) and Chi-Chi (1995). Based on the duration,peak and rise and decay era of the main pulse as well as its relationship with the predecessor and successor pulses, we propose a classification for the pulse sequences. Consequent studies have been carried out for acceleration, velocity and displacement response spectra of the main pulse with either a predecessor or a successor pulse. The analysis also includes general response behaviors in different fundamental period segments and special aspects of response at certain points (e.g., the corresponding peak points).     

基于谱修正方法的非高斯风场模拟

为克服基于H erm ite谱修正方法的缺点,减少该方法中计算H erm ite多项式系数所需耗费的大量机时,提出了一种模拟非高斯风压场的新方法,采用非高斯累积分布函数（CDF）映射技术来代替基于H erm ite的概率密度函数（PDF）修正。选择任意边缘PDF模型作为概率目标模型,采用目标功率谱密度（PSD）作为样本函数,通过迭代修正该样本函数,使其收敛于目标概率密度函数和目标功率谱密度。将该方法应用于实际结构的非高斯风场模拟,模拟结果与目标谱符合良好,表明本文方法模拟非高斯风场具有较高的精确度和计算效率。     

空间数据库离线操纵技术研究及应用

根据国家1∶50 000地形图制图数据生产基于数据库的集中式制图管理、分布式分幅式制图生产技术作业模式的实际需求,提出并设计了空间数据库的离线操纵技术,解决了数据库驱动的技术前提和分布式作业的生产现状二者之间的巨大矛盾.     

Cardboard monuments: City, language and 'nation' in contemporary Telugu cinema

This paper looks at how the city of Hyderabad, the capital of the south Indian state of Andhra Pradesh and production centre of the Telugu film industry, has been represented in recent popular cinema. It outlines the history of a new mode of representing the city in Telugu cinema and argues that its significance lies in the tendency to delocalize the city. Criminalization of the city's older inhabitants, who are marked by either religion (Islam) or their 'non-standard' dialect of Telugu, often accompanies this move. In the process, any claims that they might have on the city are delegitimized. Okkadu, a major commercial hit, restaged the country versus city dichotomy and the antagonisms complicated by cultural and political tensions between the three constituent regions of Andhra Pradesh. This film follows the late 1990s trend in the film industry to recreate entire cityscapes within the studio, ensuring that location shooting in busy city streets and neighbourhoods merely returns us to the grandeur of lavish and 'realistic' studio sets. It reconstructs Hyderabad's most recognizable monument, the 400-year-old Charminar, in addition to its obviously imaginary residential neighbourhood. This paper looks at how and why the city of Hyderabad, especially its older parts, for which Charminar is a metonym, is rendered into a fantasy space in the film.     

一种启发式有环河系自动分级算法

在地图综合中,河流分级是对水系要素表达的重点也是技术难点,对精确性和计算速度都有较高的要求。目前对河系分级的研究中,已有分级算法的效率较低,且鲜有针对有环河系的分级。基于此,本文建立了河系分级的属性规则、几何规则、环规则,对河系中存在的不同环类型进行了分类处理,且在求取河流最长路径时,运用启发式算法迭代计算,实现了河系的自动分级。实际河系数据试验表明,该算法能在对有环河系的处理中取得良好的效果,提高了河系分级的计算效率和准确度。     

Placing of photos on the internet: Critical analysis of biases on the depictions of France and Afghanistan on FLICKR

The Internet and the representation of space therein are almost omnipresent in society and everyday life. Peer-produced geographic data is gaining a particular importance through increasingly available digital tools and techniques that shape the perception of space in the internet, such as flickr, OpenStreetMap or Wikipedia. However, few studies focused on how space is represented, and by whom it is described. We hypothesize that the alleged opening up of geographic information and the assumed benefits for every individual and society through the occurrence of ‘easy-to-use-mapping-tools’ was premature. To explore these assumptions, a comparative study of the flickr worldmap was undertaken and roughly 6.8 million metadatasets of geocoded photos in France, and roughly 50,000 metadatasets in Afghanistan were downloaded and the metadata was analyzed. Our results indicate that photos geocoded in France show a large diversity of motives, while photos geocoded in Afghanistan are mostly limited to content containing warfare when they are up loaded in English. The content of the photo and therefore the representation of space strongly depend on who uploaded the photo, particularly in Afghanistan. We can show that the representation of space on the internet, for the case of flickr, is strongly dominated by perceptions of Western societies and individuals. We therefore confirm our hypothesis that the supposed opening up of geographic information systems through ‘easy-to-use-mapping-tools’ and their democratization thereof was premature. Moreover, we highlight the importance of understanding who contributes online content to be able to evaluate peer-produced data, its value, and its possible applications to avoid reproducing biases.