面向对象的湿地景观遥感分类——以杭州湾南岸地区为例

在ENVIEX软件的Feature Extraction平台上,利用LandsatTM影像数据,采用面向对象方法对杭州湾南岸地区湿地景观进行遥感影像分类;通过与基于最大似然法、人工神经网络法、支持向量机法等传统像元方法的相应分类结果进行比较,系统分析了面向对象方法在中低分辨率遥感影像的湿地景观生态分类中的有效性。研究结果表明:①较之单一依据像元光谱值进行分类的传统方法,面向对象方法综合考虑了对象的光谱、空间、纹理、色彩等多种属性特征,因而对于类型复杂多样、分布界限模糊、光谱混淆与混合像元现象严重的沿海滩涂、湖泊、河流等湿地景观具有更好的鉴别能力,也因此获得更高的分类精度(研究区景观分类总精度为88.80%,Kappa系数为0.8765);②面向对象方法在分类中提取的是由同质性像元组成的"对象",且在合理的影像分割下得到的对象破碎化程度较低,因而能在较大程度上减小分类结果中的"椒盐噪声"干扰;而基于像元方法提取的景观类型以离散像元形式组成,难以清晰表征景观的边界、形状等信息,所以分类结果中会有明显的噪声现象;③影像分割在运用面向对象方法进行遥感影像分类过程中具有重要影响,实验结果表明,60%的分割尺度和归并尺度组合较有利于中低分辨率影像的遥感分类;④面向对象分类过程中诸如影像分割精度的评价、最优分割尺度的选取、特征空间的优化等问题,则有待今后进一步探讨。     

基于低空无人机的草原灌丛遥感辨识方法

以位于中国科学院内蒙古草原生态研究定位站灌丛化样地实验平台为研究区,基于低空无人机遥感影像,结合实地调查,开展草原灌丛遥感辨识方法研究。通过对灌丛、草地和裸地归一化植被指数（NDVI）的方差统计分析,确定了裸地与植被的分割阈值为-0.08,并使用该阈值提取植被覆盖区,然后分别利用面向对象的决策树（DT）、贝叶斯（Bayes）、K最邻近（KNN）、支持向量机（SVM）机器学习分类器进行灌丛辨识。研究表明：借助Estimation of Scale Parameter（ESP）最优分割尺度评价工具可以快速确定分割参数,获取灌丛、草地影像对象;利用特征空间优化工具选取了18个的对象特征,可以有效避免盲目选择而导致的计算量增大;通过对不同分类器分类结果的对比和样本数量敏感性实验得出：Bayes分类器精度稳定、无需设置参数,灌丛分类精度最高,总体精度和Kappa系数分别达到92%和0.83,结果与影像地物嵌合最好,能够精确识别单株灌丛;根据Bayes分类器分类结果统计得研究区灌丛盖度为14.74%,平均冠幅为0.6 m²,与样方调查结果基本一致。由于4种分类器的算法特征以及对训练样本数量的敏感性各不相同,因此选择合适的分类器还需根据具体影像的地物特征、空间分辨率和研究区范围来确定。     

城市植被尺度鉴别与分类研究

城市地物具有多尺度分布特点,尺度鉴别与确定是分类的前提。提出改进的面积相对差指标,根据城市植被的分布状态确定最优分割尺度。采用面向对象方法,利用对象的光谱和空间信息对高空间分辨率影像进行植被分类。与基于像元的传统光谱分类方法和单尺度分类结果比较,最优分割尺度的鉴别和面向对象的分类方法分类精度较高,6种城市植被的分类总精度达85.5%,Kappa系数为0.83;同时有效抑制了光谱数据分类中存在的地物破碎问题。     

基于区域合并影像分割技术的多尺度地表景观分析

分割是面向对象影像分析的前提,景观空间异质性特征给遥感信息获取提出了多尺度的要求。影像分割的多尺度分析以尺度效应与对象异质性最小原则为前提,描述多尺度效应分析的必要性与可行性。基于区域合并的分割原理与方法,为多尺度影像分析提供理论与技术支持,在两个样区进行多尺度影像分割技术的应用实践,分析基于区域合并的多尺度分割数据的几何丰富度与语义丰富度。     

Human population distribution modelling at regional level using very high resolution satellite imagery

Modelling the distribution of human population based on satellite-derived information has become an important field of research, providing valuable input e.g. for human impact assessments related to the management of threatened ecosystems. However, few regional-scale studies have been conducted in developing countries, where detailed land cover data is usually absent, and the potential of very high resolution (VHR) satellite imagery in this context has not been explored yet. This study uses results obtained through object-based image analysis (OBIA) of QuickBird imagery for a subset of a highly populated rural area in western Kenya. Functions are established that approximate frequency distributions of QuickBird-derived locations of houses in relation to five factors. These factors are known to impact settlement patterns and data is available for the entire study area. Based on an overall probability coefficient (weight) calculated from the single functions, human population is redistributed at the smallest administrative level available (version A). In addition, the problem of artefacts remaining at administrative boundaries is addressed by combining the approach with the pycnophylactic smoothing algorithm (Tobler, 1979) (version B). The results show distinct patterns of population distribution, with particular influence of rivers/streams and slope, while version B in addition is free of boundary artefacts. Despite some limitations compared to models based on detailed land cover data (e.g. the ability of capturing abrupt changes in population density), a visual and numerical evaluation of the results shows that using houses as classified from VHR imagery for a study area subset works well for redistributing human population at the regional level. This approach might be suitable to be applied also in other regions of e.g. sub-Saharan Africa.     

Increasing the spatial resolution of agricultural land cover maps using a Hopfield neural network

Land cover class composition of remotely sensed image pixels can be estimated using soft classification techniques increasingly available in many GIS packages. However, their output provides no indication of how such classes are distributed spatially within the instantaneous field of view represented by the pixel. Techniques that attempt to provide an improved spatial representation of land cover have been developed, but not tested on the difficult task of mapping from real satellite imagery. The authors investigated the use of a Hopfield neural network technique to map the spatial distributions of classes reliably using information of pixel composition determined from soft classification previously. The approach involved designing the energy function to produce a ‘best guess’ prediction of the spatial distribution of class components in each pixel. In previous studies, the authors described the application of the technique to target identification, pattern prediction and land cover mapping at the sub-pixel scale, but only for simulated imagery. We now show how the approach can be applied to Landsat Thematic Mapper (TM) agriculture imagery to derive accurate estimates of land cover and reduce the uncertainty inherent in such imagery. The technique was applied to Landsat TM imagery of small-scale agriculture in Greece and largescale agriculture near Leicester, UK. The resultant maps provided an accurate and improved representation of the land covers studied, with RMS errors for the Landsat imagery of the order of 0.1 in the new fine resolution map recorded. The results showed that the neural network represents a simple efficient tool for mapping land cover from operational satellite sensor imagery and can deliver requisite results and improvements over traditional techniques for the GIS analysis of practical remotely sensed imagery at the sub pixel scale.     

Extracting ecological information from oblique angle terrestrial landscape photographs: Performance evaluation of the WSL Monoplotting Tool

While aerial photography and satellite imagery are the usual data sources used in remote sensing, land based oblique photographs can also be used to measure ecological change. By using such historical photographs, the time frame for change detection can be extended into the late 1800s and early 1900s, predating the era of aerial imagery by decades. Recent advancements in computing power have enabled the development of techniques for georeferencing oblique angle photographs. The WSL Monoplotting Tool is a new piece of software that opens the door to analyzing such photographs by allowing for extraction of spatially referenced vector data from oblique photographs. A very large repeat photography collection based on the world's largest systematic collection of historical mountain topographic survey images, the Mountain Legacy Project, contains >6000 high resolution oblique image pairs showing landscape changes in the Rocky Mountains of Alberta between ca. 1900 – today. We used a subset of photographs from this collection to assess the accuracy and utility of the WSL Monoplotting Tool for georeferencing oblique photographs and measuring landscape change. We determined that the tool georeferenced objects to within less than 15 m of their real world 3D spatial location, and the displacement of the geographic center of over 121 control points was less than 3 m from the real world spatial location. Most of the error in individual object placement was due to the angle of viewing incidence with the ground (i.e., low angle/highly oblique angles resulted in greater horizontal error). Simple rules of control point selection are proposed to reduce georeferencing errors. We further demonstrate a method by which raster data can be rapidly extracted from an image pair to measure changes in vegetation cover over time. This new process permits the rapid evaluation of a large number of images to facilitate landscape scale analysis of oblique imagery.     

A land use/land cover change geospatial cyberinfrastructure to integrate big data and temporal topology

ABSTRACT

Big data have shifted spatial optimization from a purely computational-intensive problem to a data-intensive challenge. This is especially the case for spatiotemporal (ST) land use/land cover change (LUCC) research. In addition to greater variety, for example, from sensing platforms, big data offer datasets at higher spatial and temporal resolutions; these new offerings require new methods to optimize data handling and analysis. We propose a LUCC-based geospatial cyberinfrastructure (GCI) that optimizes big data handling and analysis, in this case with raster data. The GCI provides three levels of optimization. First, we employ spatial optimization with graph-based image segmentation. Second, we propose ST Atom Model to temporally optimize the image segments for LUCC. At last, the first two domain ST optimizations are supported by the computational optimization for big data analysis. The evaluation is conducted using DMTI (DMTI Spatial Inc.) Satellite StreetView imagery datasets acquired for the Greater Montreal area, Canada in 2006, 2009, and 2012 (534 GB, 60 cm spatial resolution, RGB image). Our LUCC-based GCI builds an optimization bridge among LUCC, ST modelling, and big data.     

WyomingView: No-Cost Remotely Sensed Data for Geographic Education

Abstract

Learning enhanced by visual examples and remotely sensed imagery is a valuable classroom resource for teaching students geographic concepts in a meaningful context. Barriers to the use of imagery include difficulty finding appropriate imagery and the cost of moderate resolution satellite imagery. A program in Wyoming called WyomingView and analogous programs in other states are providing no-cost, preprocessed satellite imagery delivered over the Internet that can help teachers better communicate geospatial knowledge to their students.     

农田景观空间异质性分析及遥感监测最优尺度选择——以三江平原为例

农情遥感监测需要高时间分辨率的遥感数据,目前这些数据大都为中低空间分辨率影像。在这些尺度下,像元内部往往是异质的,从而影响农情参数反演精度。因此分析和表达农田景观空间异质性和最优尺度选择对遥感农情监测质量的提高具有重要的应用价值。选取建三江农垦区四种典型农田景观为研究点,Landsat/TM NDVI为实验数据,利用实验变异函数对四种景观类型的各向空间异质性进行了分析, 而后通过变异函数模型拟合,定量分析了各个研究点的整体空间异质性,并在此基础上进行了研究区遥感监测最优尺度选择。研究表明:(1) 基于实验变异函数的结构分析方法,可定性地认识空间异质性的大小和方向,进而挖掘出其背后的自然和人为驱动因素。(2) 对实验变异函数进行拟合分析,可定量地刻画不同景观格局各自的空间异质性特性。此外,基于变异函数对空间异质性的定量表达,讨论了利用积分变程A结合Nyquist-Shannon采样定理进行最优尺度选择的方法。     

Supporting local health decision making with spatial video: Dengue,Chikungunya and Zika risks in a data poor,informal community in Nicaragua

One challenge facing spatial scientists trying to support public health outreach and intervention in challenging environments is the lack of fine scale spatial data. These data are required to gain a better understanding of both physical and social systems; why disease occurs where it does, and how to disrupt it. While data options exist, including high resolution aerial imagery, remotely sensed data, and even online mapping products like Google Street View, these all come with limitations. One option that has previously been utilized to assess cholera risk is spatial video. Here it is used to map potential mosquito breeding sites in an endemic Dengue and Chikungunya, and emerging Zika impacted community. We show how this method can provide mapping support in the hands of non-specialist public health workers who, working in collaboration with out-of-area geographic information systems (GIS) teams, can identify where to target limited intervention resources. We use a case study of an impoverished informal style Nicaraguan community suffering from a high disease burden to show spatial variation in potential mosquito breeding habitats. A field team collected street-by-street spatial video data to produce fine scale risk maps of standing water and trash locations, which, when interpreted with the associated spatial video imagery, were used to suggest where intervention strategies should be targeted. We also discuss how these same data layers can be used to address other health concerns traditionally found in informal settlements.     

结合多尺度纹理的高分辨率遥感影像决策树分类

地物具有多尺度特点,遥感影像包含的地物纹理信息很难用单一尺度来描述。通过选择最佳纹理尺度组合,利用光谱数据结合多尺度纹理对高分辨率影像进行决策树分类。研究结果表明:结合多尺度纹理的高分辨遥感影像决策树分类,能够更好地描述地物并有效解决光谱数据分类中存在的地物破碎问题,其分类精度为81.7%,kap-pa系数为0.78;与光谱数据分类和结合单尺度纹理数据分类结果比较,分类精度分别提高了11.2%和6%,该方法有助于提高高分辨率影像的分类精度。     

地理学视角下土地利用/覆被分类发展探讨

随着对地观测和互联网技术的发展,地理大数据时代正在到来,其多尺度、长时序、多模态等海量“超”覆盖数据为土地利用/覆被（Land Use/Land Cover, 简称LULC）分类及变化检测带来巨大的机遇,支撑着新时代人、地两大系统相互作用关系的认知和实践。然而,多数地理学者认为地理学基本原理与核心思想并未因为大数据的到来而发生本质性变化。所以,从地理学基本原理角度理解LULC分类的发展,尤其在地理大数据时代的发展方向,不失为一条可行的途径。为此,本文从区域、尺度、综合三方面的地理学基本原理视角将LULC分类技术的发展划分为地球观测数据匮乏阶段、人类行为数据融合阶段以及地理大数据“超”覆盖阶段分别探讨分析,以期主动把握LULC分类技术及应用的未来发展趋势。研究结果显示：在地球观测数据匮乏阶段,LULC分类多以类型还不丰富的遥感数据源,在空间分辨率较低的像元尺度上,进行以地表覆被状态为主的分类;发展到人类行为数据融合阶段,LULC分类在城市区域率先出现了对地观测数据和人类行为数据相融合,在街区尺度上进行以空间功能异质性划分、识别为主导的城市功能区分类;在地理大数据“超”覆盖阶段,LULC分类将实现多尺度协同、面向全空间的功能异质性划分,并在主体功能的基础上融合“社会-经济-自然”多维定量属性,本文称之为“空间场景”。希望本文的探讨能够为地理大数据时代LULC分类的新技术发展和新产品应用提供有益启示。     

Image data fusion for the remote sensing of freshwater environments

Remote sensing based mapping of diverse and heterogeneous freshwater environments requires high-resolution images. Data fusion is a useful technique for producing a high-resolution multispectral image from the merging of a high-resolution panchromatic image with a low-resolution multispectral image. Given the increasing availability of images from different satellite sensors that have different spectral and spatial resolutions, data fusion techniques that combine the strengths of different images will be increasingly important to Geography for land-cover mapping. Different data fusion methods however, add spectral and spatial distortions to the resultant data depending on the geographical context; therefore a careful selection of the fusion method is required. This paper compares a technique called subtractive resolution merge, which has not previously been formally tested, with conventional techniques such as Brovey transformation, principal component substitution, local mean and variance matching, and optimised high pass filter addition. Data fusion techniques are grouped into spectral and spatial centric methods. Subtractive resolution merge belongs to a new class of data fusion techniques that uses a mix of both spatial and spectral centric approaches. The different data fusion techniques were applied to a QuickBird image of a semi-aquatic freshwater environment in New Zealand. The results were compared both qualitatively and quantitatively using spectral and spatial error metrics. This research concludes that subtractive resolution merge performed better than all the other techniques and will be a valuable technique for enhancing images for freshwater land-cover mapping.     

面向对象的山地区域多源遥感影像分割尺度选择及评价

随着遥感技术的飞速发展,通过各种传感器获取的影像种类越来越多.传统的面向像元的影像分析技术已难以适应高分辨率影像,面向对象的影像分析技术应运而生.面向对象的影像分析技术所面临的首要问题是影像分割,如何选择合适的影像分割尺度已成为面向对象影像分析技术的关键.为了提高遥感影像分割尺度选择的准确性,通过对影像分析建立了分割质量函数,针对不同分辨率的多源遥感影像进行分割实验,获取了最优分割尺度,最后提出了一种顾及面积和光谱的分割匹配指数对最优分割尺度进行评价,并研究了分割尺度与亮度均值标准差、影像空间分辨率之间的对应关系.研究结果表明,构建的分割质量函数获取的最优分割尺度避免了人工分割带来的主观性,建立的同时顾及面积和光谱因子的分割匹配指数验证了最优分割尺度选择的正确性.     

基于差值权重和快速傅立叶变换的“北京一号”卫星影像融合方法研究

提出一种基于差值权重和快速傅立叶变换的影像融合方法--DWF,对"北京一号"小卫星的全色和多光谱影像进行融合.该算法根据卫星影像的成像过程,通过傅立叶变换获取全色影像的高频、低频信息和多光谱影像在HIS颜色空间中分量的低频信息,由这3种信息通过基于差值权重的决策计算获得融合影像的空间信息.实验采用PCA、GS、DWF以及基于小波的HIS变换的融合方法,结果表明DWF方法优于其他融合方法,比较适合于"北京一号"卫星的全色与多光谱影像融合.     

Scale- and orientation-invariant scene similarity metrics for image queries

In this paper we extend our previous work on shape-based queries to support queries on configurations of image objects. Here we consider spatial reasoning, especially directional and metric object relationships. Existing models for spatial reasoning tend to rely on pre-identified cardinal directions and minimal scale variations, assumptions that cannot be considered as given in our image applications, where orientations and scale may vary substantially, and are often unknown. Accordingly, we have developed the method of varying baselines to identify similarities in direction and distance relations. Our method allows us to evaluate directional similarities without a priori knowledge of cardinal directions, and to compare distance relations even when query scene and database content differ in scale by unknown amounts. We use our method to evaluate similarity between a user-defined query scene and object configurations. Here we present this new method, and discuss its role within a broader image retrieval framework.     

沈阳城市应急避难场所的识别、空间格局与服务功能

以沈阳市为例,运用高分辨率遥感影像识别城市应急避难场所,采用空间插值、波段集统计等方法刻画应急避难场所空间分布及其与人口的空间匹配性,通过栅格成本分析和加权Voronoi图等方法测算空间可达性并划分其服务区范围。研究表明:(1)沈阳市应急避难场所分布呈现出"小集聚、大分散"的基本格局,与人口分布的空间匹配性差;(2)应急避难场所可达性表现出空间异质性,大型用地对整体格局起到主要塑造作用;(3)研究区域可划分为14个较为明确的应急避难场所服务区,不同服务范围在空间上存在大小、数量和构成上的差异。最后,从城市安全视角对沈阳市应急避难场所的空间布局提出优化建议,为政府部门决策提供科学依据与政策参考。     

基于阴影搜索法的飞机目标遥感图像分割研究

介绍阈值分割的原理并分析传统图像分割中阈值的确定方法;为提高对飞机目标遥感图像的分割效果,采用阴影搜索法对图像阈值进行判定。该算法对图像进行横向或纵向扫描,根据像素的变化频率判定阴影区域并确定目标与背景的关系,实现图像分割中最优阈值的自动提取。该算法克服了传统方法灵敏度低对分割效果的影响,有很好的通用性和有效性。     

APPLICATIONS OF ARTIFICIAL INTELLIGENCE TECHNIQUES TO REMOTE SENSING*

Remote sensing is a geographic analysis tool capable of producing large quantities of data in the spectral, temporal and spatial domains. Techniques for automating the image analysis process would be advanced by the inclusion of artificial intelligence techniques in the design of image processing systems. The remote sensing application which show promise for successful implementation of artificial intelligence techniques are intelligent onboard processing, advanced database interrogation, and the automated analysis of multispectral imagery.