多分类器融合的遥感影像分类方法实验研究

在遥感影像分类应用中,使用不同种类的分类器对原始图像进行操作所得出的分类结果的分类精度和效率都是不同的,设计完成多分类器融合分类实验,实验可以将不同种单一分类器的优点用适当的方法进行组合,从而获得比单一分类器分类精度和效率更好的分类方法.在此基础上采用投票原理设计抽象级融合的多分类器并完成实验.设计结果表明:多分类器融合的遥感影像分类方法在精度上要比单一分类器的分类精度高.      

多源数据的面向对象国际河流土地覆被分类研究

作为横跨3个国家(尼泊尔、印度、中国)的国际跨界河流———柯西河流域,地形高差巨大,土地覆被结构组成复杂,进行土地覆被的自动分类研究具有典型意义。基于面向对象方法多源遥感数据、训练规则、丰富的细节信息为复杂土地覆被自动分类研究提供了可能。选择合适的影像分割特征和最优分割尺度,按照数据挖掘中的规则顺序逐步进行各个土地覆被的提取。总体精度说明分类结果与野外点相一致的概率能达到90.05%,说明国际跨界河流土地覆被分类方法是可行的,分类结果是准确、可信的。     

NWFE结合纹理特征的SVM土地覆被分类方法研究

为提高土地覆被分类精度,采用非参数权重特征提取(nonparametric weighted feature extraction,NWFE)结合纹理特征的支持向量机(support vector machines,SVM)的分类法,对新疆玛纳斯河流域绿洲区2006年的土地覆被进行分类,并将该方法与主成分分析(principal component analysis,PCA)结合纹理特征的SVM分类、原始波段结合纹理特征的SVM分类进行对比。结果表明,NWFE结合纹理特征的SVM分类结果优于其他2种分类结果,不仅反映了土地覆被分布的整体情况,而且使不同土地覆被类型得到较好的区分,总体分类精度达89.17%。     

粗集属性划分的集成遥感分类

提出了一种基于粗集属性划分的遥感分类新方法, 构造了基于粗集的集成遥感分类器。该分类器利用粗集理论将输入的属性集合划分为多个约减, 利用这些约减构造多个训练子集。每个训练子集训练神经网分类器, 在决策时将多个单个分类器的结果进行投票选举。这种方法即减少了单个分类器的输入属性个数, 又避免了由于属性选取造成单一分类器在某些分类上的错误偏见。该分类器与神经网分类器方法, 以及属性选取与神经网结合方法进行了比较。结果表明RSEC无论在分类精度上, 还是在不同样本个数条件下的精度稳定程度上均有较好表现。     

基于面向对象与规则的Sentinel-2A影像土地覆被分类——以江西省都昌县为例

土地覆被作为地表自然和人工建造物的综合体，是开展土地科学相关研究的重要基础，在遥感大数据背景下，准确、快速、自动化进行土地覆被提取技术一直是遥感研究中的重点。本文基于eCognition软件，采用面向对象的多尺度分割法，综合考虑地物在遥感影像上的光谱、形状和纹理特征，建立多种地物提取规则。通过模糊函数、支持向量机（SVM）和阈值法对研究区的土地覆被进行分类提取，并与研究区的FROM-GLC10数据和土地利用变更数据进行了对比分析。结果表明：①研究区土地覆被分类的总体精度为97%，Kappa系数为0.96，分类精度较高；②基于10 m分辨率影像，综合使用形状、纹理、光谱信息对于道路的提取具有较好的效果，道路提取Kappa系数为0.84；③分类结果在面积和空间分布上都优于FROM-GLC10数据，与研究区实际土地变更数据保持较好的一致性。基于面向对象与规则的分类方法提取地物能够有效利用多种遥感影像特征，分类精度高，对于处理高分辨率遥感数据具有很好的优势。     

基于多尺度影像分割的面向对象城市土地覆被分类研究——以马来西亚吉隆坡市城市中心区为例

城市受人类活动影响比较大,结构组成比较复杂,对该区域进行分类研究存在一些问题。甚高分辨率遥感影像,以其丰富的细节信息为城市土地覆被分类研究提供了可能。本文结合使用甚高分辨率QuickBird遥感影像和激光扫描LIDAR数据,论述了利用多尺度、多变量影像分割的面向对象的分类技术对马来西亚基隆坡市城市中心区的土地覆被分类研究。针对特定地物选择合适的影像分割特征和分割尺度、按照合理的提取顺序逐步进行城市土地覆被信息提取。在建筑物的提取过程中构建了归一化数字表面模型nDSM,使用成员函数将建筑物信息提取出来。精度评价结果表明,利用该方法得到了理想的城市土地覆被分类结果,其分类总精度从常规面向对象分类方法的83.04%上升到88.52%,其中建筑物生产精度从60.27%增加到93.91%。     

基于BP神经网络和MODIS-EVI时间序列的土地覆被分类

以matlab为平台设计了BP神经网络用于自动提取土地覆被信息,将覆盖内蒙古的2006年全年的MO- DIS数据处理成EVI植被指数时间序列集输入神经网络进行土地覆被分类,精度检验结果显示:总体分类精度(o- verall-accuracy)为78.32%,kappa系数为0.73。     

多源数据的土地覆被样本自动提取

随着遥感数据获取能力的不断增强,自动化程度已经成为大尺度遥感土地覆被分类面临的关键问题。然而,现有训练样本的人工选取方法成为制约土地覆被分类自动化的瓶颈。本文以河南、贵州两省为研究区,提出一种基于多源数据的土地覆被样本自动提取方法,以构建适用于大尺度的土地覆被自动分类。首先,以2010年1∶10万土地利用数据CHINALC和30 m分辨率全球土地覆被数据Globle Land30为样本数据源;然后,利用空间一致性分析及异质性分析确定样本初选区域;最后,通过样本提纯去除无效样本。结果表明:(1)应用多源数据的土地覆被样本自动提取方法获得的分类产品总体分类精度高于人工样本提取方法制作的全球土地覆被产品MCD12Q1。(2)与单源样本自动提取方法相比,应用多源数据的土地覆被样本自动提取方法,可获得更好的分类稳定性。综上,多源数据的土地覆被样本自动提取方法可在保证精度的同时,提升土地覆被分类的自动化程度。     

多源遥感技术在土地利用分类中的应用

为有效提高土地利用分类精度,研究多源遥感数据的分类方法,提出了利用多源遥感技术进行土地利用分类。对居民地、耕地、林地、水体、未利用土地等土地利用类型的光谱特征及微波散射特征进行了分析,将高分一号多光谱数据与RADARSAT-2数据相结合,利用决策树分类器实现了土地利用类型的划分,其总体分类精度达到96.6%。与高分一号数据最大似然分类及RADARSAT-2数据的Freeman-Durden三分量最大似然分类进行了精度比较,结果表明,多源遥感技术可实现数据的特征互补,其分类精度优于仅采用多光谱数据或微波数据的分类精度。     

多分类器联合监督分类方法研究

针对遥感图像监督分类方法适用范围不同且分类机制各有优劣的特点,本文提出将最大似然法与最小距离法结合的监督分类法。对eTM 影像进行分类,结果表明,与单一分类器的分类结果相比,分类器结合的监督分类技术能有效提高遥感图像专题信息提取的精度。     

基于误差分析的组合分类器研究

提出了一种基于误差分析的组合分类器,通过结合两种监督分类方法,提出的算法分别估计了两种监督分类方法在计算过程中的误差,给出了规则输出的置信区间,再根据置信区间的大小对两种分类方法的输出结果进行加权平均,从而得到更精确的规则输出.利用该方法对遥感图像进行分类实验,在不同训练样本分布与不同训练样本数量的情况下,比较新的组合分类器与单一分类器的精度.结果表明新的组合分类器能够取得比单一的分类器更高的分类精度.结果还显示出,两个分类器的独立性越强,组合分类器的效果越好.另外一个实验比较了新的组合分类器与和式规则组合分类器的分类精度,结果仍显示出了新方法的优越性.     

Enhanced land use/cover classification of heterogeneous tropical landscapes using support vector machines and textural homogeneity

Land use/cover classification is a key research field in remote sensing and land change science as thematic maps derived from remotely sensed data have become the basis for analyzing many socio-ecological issues. However, land use/cover classification remains a difficult task and it is especially challenging in heterogeneous tropical landscapes where nonetheless such maps are of great importance. The present study aims at establishing an efficient classification approach to accurately map all broad land use/cover classes in a large, heterogeneous tropical area, as a basis for further studies (e.g., land use/cover change, deforestation and forest degradation). Specifically, we first compare the performance of parametric (maximum likelihood), non-parametric (k-nearest neighbor and four different support vector machines – SVM), and hybrid (unsupervised–supervised) classifiers, using hard and soft (fuzzy) accuracy assessments. We then assess, using the maximum likelihood algorithm, what textural indices from the gray-level co-occurrence matrix lead to greater classification improvements at the spatial resolution of Landsat imagery (30 m), and rank them accordingly. Finally, we use the textural index that provides the most accurate classification results to evaluate whether its usefulness varies significantly with the classifier used. We classified imagery corresponding to dry and wet seasons and found that SVM classifiers outperformed all the rest. We also found that the use of some textural indices, but particularly homogeneity and entropy, can significantly improve classifications. We focused on the use of the homogeneity index, which has so far been neglected in land use/cover classification efforts, and found that this index along with reflectance bands significantly increased the overall accuracy of all the classifiers, but particularly of SVM. We observed that improvements in producer's and user's accuracies through the inclusion of homogeneity were different depending on land use/cover classes. Early-growth/degraded forests, pastures, grasslands and savanna were the classes most improved, especially with the SVM radial basis function and SVM sigmoid classifiers, though with both classifiers all land use/cover classes were mapped with producer's and user's accuracies of ∼90%. Our classification approach seems very well suited to accurately map land use/cover of heterogeneous landscapes, thus having great potential to contribute to climate change mitigation schemes, conservation initiatives, and the design of management plans and rural development policies.     

A kernel functions analysis for support vector machines for land cover classification

Information about the Earth's surface is required in many wide-scale applications. Land cover/use classification using remotely sensed images is one of the most common applications in remote sensing, and many algorithms have been developed and applied for this purpose in the literature. Support vector machines (SVMs) are a group of supervised classification algorithms that have been recently used in the remote sensing field. The classification accuracy produced by SVMs may show variation depending on the choice of the kernel function and its parameters. In this study, SVMs were used for land cover classification of Gebze district of Turkey using Landsat ETM+ and Terra ASTER images. Polynomial and radial basis kernel functions with their estimated optimum parameters were applied for the classification of the data sets and the results were analyzed thoroughly. Results showed that SVMs, especially with the use of radial basis function kernel, outperform the maximum likelihood classifier in terms of overall and individual class accuracies. Some important findings were also obtained concerning the changes in land use/cover in the study area. This study verifies the effectiveness and robustness of SVMs in the classification of remotely sensed images.     

全球地表覆盖制图研究新范式

全球土地覆盖制图在过去的10年中取得重要进展,空间分辨率从300 m增加至30 m,分类详细程度也有所提高,从10余个一级类到包含29类的二级分类体系。然而,利用光学遥感数据在大空间范围制图方面仍有诸多挑战。本文主要介绍在农田、居住区、水体和湿地制图方面的挑战,讨论在使用多时相和多传感器遥感数据上的困难,这将是未来遥感应用的趋势。由于各种地表覆盖数据产品有自己定义的地表覆盖类型体系和处理流程,通过调和以及集成各种全球土地覆盖制图产品能够满足新的应用目的,并且可以最大程度地利用已有的土地覆盖数据。然而,未来全球土地覆盖制图需要能够按照新应用需求动态生成地表覆盖数据产品的能力。过去的研究表明有效地提高局部尺度制图的分类精度,更好的算法、更多种特征变量(新类型的数据或特征)以及更具代表性的训练样本都非常重要。我们却认为特征变量的使用更重要。本文提出了一个全球土地覆盖制图的新范式。在这个新范式中,地表覆盖类型的定义被分解为定性指标的类、定量指标的植被郁闭度和高度。非植被类型通过它们的光谱和纹理信息提取。复合考虑类、郁闭度和高度3种指标来定义和区别包含植被的地表覆盖类型。郁闭度和高度不能在分类算法中提取,需要借助其他直接测量或间接反演方法。新的范式还表明,一个普遍适用的训练样本集有效地提高了在非洲大陆尺度土地覆盖分类。为了确保更加容易地实现从传统的土地覆盖制图到全球土地覆盖制图新范式的转变,建议构建一体化的数据管理和分析系统。通过集成相关的观测数据、样本数据和分析算法,逐步建成全球土地覆盖制图在线系统,构建全球地表覆盖制图门户网站,为数据生产者、数据用户、专业研究人员、决策人员搭建合作互助的平台。     

Soft classification for sub-pixel land cover extraction

With increasing resolution of the remotely sensed data the problems of images contaminated by mixed pixels arc frequent. Conventional classification techniques often produce erroneous results when applied to images dominated by mixed pixels. This may load to unrealistic representation of land cover, thereby, affecting efficient planning, management and monitoring of natural resources. Consequently, soft classification techniques providing sub-pixel land cover information may have to be utilised. From a range of soft classification techniques, the present study focuses on the utility of conventional maximum likelihood classifier and linear mixture modelling for sub-pixel. land cover classifications. The accuracy of the soft classifications has been assessed using distance measures and correlation co-efficient. The results show that linear mixture modelling has produced accuracies comparable to maximum likelihood classifier. Besides this the correlations between actual land cover proportions and proportions from linear mixture modelling, though not strong, arc statistically significant at 95% level of confidence. It has also been observed that the normalised likelihoods of maximum likelihood classifier also show strong correlations with the actual land cover proportions on ground and therefore has the potential to be used as a soft classification technique.     

Assessing reference dataset representativeness through confidence metrics based on information density

Land cover maps obtained from classification of remotely sensed imagery provide valuable information in numerous environmental monitoring and modeling tasks. However, many uncertainties and errors can directly or indirectly affect the quality of derived maps. This work focuses on one key aspect of the supervised classification process of remotely sensed imagery: the quality of the reference dataset used to develop a classifier. More specifically, the representative power of the reference dataset is assessed by contrasting it with the full dataset (e.g. entire image) needing classification. Our method is applicable in several ways: training or testing datasets (extracted from the reference dataset) can be compared with the full dataset. The proposed method moves beyond spatial sampling schemes (e.g. grid, cluster) and operates in the multidimensional feature space (e.g. spectral bands) and uses spatial statistics to compare information density of data to be classified with data used in the reference process. The working hypothesis is that higher information density, not in general but with respect to the entire classified image, expresses higher confidence in obtained results. Presented experiments establish a close link between confidence metrics and classification accuracy for a variety of image classifiers namely maximum likelihood, decision tree, Backpropagation Neural Network and Support Vector Machine. A sensitivity analysis demonstrates that spatially-continuous reference datasets (e.g. a square window) have the potential to provide similar classification confidence as typically-used spatially-random datasets. This is an important finding considering the higher acquisition costs for randomly distributed datasets. Furthermore, the method produces confidence maps that allow spatially-explicit comparison of confidence metrics within a given image for identification of over- and under-represented image portions. The current method is presented for individual image classification but, with sufficient evaluation from the remote sensing community it has the potential to become a standard for reference dataset reporting and thus allowing users to assess representativeness of reference datasets in a consistent manner across different classification tasks.     

An evaluation of Radarsat-2 individual and combined image dates for land use/cover mapping

Various land use/cover types exhibit seasonal characteristics which can be captured in remotely sensed imagery. This study examined how different seasons of Radarsat-2 data influence land use/cover classification accuracies for two study sites. Two dates of Radarsat-2 C-band quad-polarised images were obtained for Washington, DC, USA and Wad Madani, Sudan. Spectral signatures were extracted and used with a maximum likelihood decision rule for classification and thematic accuracies were then determined. Both despeckled radar and derived texture measures were examined. Thematic accuracies for the two despeckled image dates were similar with a difference of 3% for Washington and 6% for Sudan. Merging the despeckled images for both seasons increased overall accuracy by 2% for Washington and 9% for Sudan. Further combining the original radar for both seasons with derived texture measures increased overall accuracies by 9% for Washington and 16% for Sudan for final overall accuracy values of 73 and 82%.     

Classifiers vs. input variables—The drivers in image classification for land cover mapping

The study investigates the performance of image classifiers for landscape-scale land cover mapping and the relevance of ancillary data for the classification success in order to assess and to quantify the importance of these components in image classification. Specifically tested are the performance of maximum likelihood classification (MLC), artificial neural networks (ANN) and discriminant analysis (DA) based on Landsat7 ETM+ spectral data in combination with topographic measures and NDVI. ANN produced high accuracies of more than 75% also with limited input information, while MLC and DA produced comparable results only by incorporating ancillary data into the classification process. The superiority of ANN classification was less pronounced on the level of the single land cover classes. The use of ancillary data generally increased classification accuracy and showed a similar potential for increasing classification accuracy than the selection of the classifier. Therefore, a stronger focus on the development of appropriate and optimised sets of input variables is suggested. Also the definition and selection of land cover classes has shown to be crucial and not to be simply adaptable from existing land cover class schemes. A stronger research focus towards discriminating land cover classes by their typical spectral, topographic or seasonal properties is therefore suggested to advance image classification.     

Assessing the performance of two unsupervised dimensionality reduction techniques on hyperspectral APEX data for high resolution urban land-cover mapping

Despite the high richness of information content provided by airborne hyperspectral data, detailed urban land-cover mapping is still a challenging task. An important topic in hyperspectral remote sensing is the issue of high dimensionality, which is commonly addressed by dimensionality reduction techniques. While many studies focus on methodological developments in data reduction, less attention is paid to the assessment of the proposed methods in detailed urban hyperspectral land-cover mapping, using state-of-the-art image classification approaches. In this study we evaluate the potential of two unsupervised data reduction techniques, the Autoassociative Neural Network (AANN) and the BandClust method – the first a transformation based approach, the second a feature-selection based approach – for mapping of urban land cover at a high level of thematic detail, using an APEX 288-band hyperspectral dataset. Both methods were tested in combination with four state-of-the-art machine learning classifiers: Random Forest (RF), AdaBoost (ADB), the multiple layer perceptron (MLP), and support vector machines (SVM). When used in combination with a strong learner (MLP, SVM) BandClust produces classification accuracies similar to or higher than obtained with the full dataset, demonstrating the method’s capability of preserving critical spectral information, required for the classifier to successfully distinguish between the 22 urban land-cover classes defined in this study. In the AANN data reduction process, on the other hand, important spectral information seems to be compromised or lost, resulting in lower accuracies for three of the four classifiers tested. Detailed analysis of accuracies at class level confirms the superiority of the SVM/Bandclust combination for accurate urban land-cover mapping using a reduced hyperspectral dataset. This study also demonstrates the potential of the new APEX sensor data for detailed mapping of land cover in spatially and spectrally complex urban areas.     

A comparative assessment between object and pixel-based classification approaches for land use/land cover mapping using SPOT 5 imagery

Abstract

Land use/land cover (LULC) classification with high accuracy is necessary, especially in eco-environment research, urban planning, vegetation condition study and soil management. Over the last decade a number of classification algorithms have been developed for the analysis of remotely sensed data. The most notable algorithms are the object-oriented K-Nearest Neighbour (K-NN), Support Vector Machines (SVMs) and the Decision Trees (DTs) amongst many others. In this study, LULC types of Selangor area were analyzed on the basis of the classification results acquired using the pixel-based and object-based image analysis approaches. SPOT 5 satellite images with four spectral bands from 2003 and 2010 were used to carry out the image classification and ground truth data were collected from Google Earth and field trips. In pixel-based image analysis, a supervised classification was performed using the DT classifier. On the other hand, object-oriented (K-NN) image analysis was evaluated using standard nearest neighbour as classifier. Subsequently SVM object-based classification was performed. Five LULC categories were extracted and the results were compared between them. The overall classification accuracies for 2003 and 2010 showed that the object-oriented (K-NN) (90.5% and 91%) performed better results than the pixel-based DT (68.6% and 68.4%) and object-based SVM (80.6% and 78.15%). In general, the object-oriented (K-NN) performed better than both DTs and SVMs. The obtained LULC classification maps can be used to improve various applications such as change detection, urban design, environmental management and zooning.