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.     

长时间序列Radarsat图像的神经网络模拟及土地覆盖变化的快速检测

土地覆盖的短期时空变化模式研究,对土地覆盖的快速、动态监测具有重要意义,也是遥感研究的新热点。本文利用2000—2001年的时间序列Radarsat图像,采用功率谱分析方法,对土地覆盖的短期时—空变化的周期特征进行了分析,由此建立了基于时间序列影像分析的神经网络预测模型,从植被主要生长季节的时间序列雷达卫星影像获取训练样本,对研究区域的典型土地覆盖的短期动态变化过程进行了学习。学习后的模型能够利用多个时间序列的Radarsat影像对下一时刻的影像进行模拟,并进一步检测变化。在模拟结果基础上,定义相对变化距离函数和检测门限,对模拟影像及实际影像中的变化区域进行了检测。检测精度范围在66.67%(农村居民点)—91.67%(水体)之间,平均检测精度为81.66%。由于时间序列信号的引入,神经网络模型能够较好地获取土地覆盖的短期动态变化信息。     

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.     

Neural networks-based simulation of land cover scenarios in Doon valley,India

Land cover transformation is one of the foremost aspects of human-induced environmental change, having an extensive history dating back to antiquity. The present study aims to simulate the process of land cover change based on different policy-based scenarios so as to provide a basis for sustainable development in Doon valley, India. For this purpose, an artificial neural network-based spatial predictive model was developed for the Doon valley. The predictive model generated future land cover patterns under three policy scenarios, i.e. baseline scenario, compact growth scenario and hierarchical growth scenario (HGS). The simulated land cover patterns mirror where land cover patterns are headed in the valley by year 2021. The result suggests that unabated continuation of the present pattern of land cover transformation will result in a regional imbalance. However, this skewed development can be corrected by altering the current growth trend as revealed in the compact growth and HGSs.     

一种基于联合特征的地表覆盖变化自动检测方法

地表覆盖的高效变化检测在地理国情监测中具有重要意义。本文针对当前地表覆盖检测人工目视解译方法效率低,以及软件自动解译错检率、漏检率较高的特点和现状,提出了一种基于联合特征的地表覆盖类型自动变化检测方法。该方法通过对比7种不同的特征联合方案,确立了联合灰度共生矩阵、灰度直方图、光谱统计特征、对象特征的最优组合形式,并设计支持向量机高维度分类器进行分类。试验结果表明,在浙江省复杂地表覆盖分布情况下,基于分辨率优于1 m的国产高分卫星影像,该方法对房屋建筑区、建筑工地等人工构筑物类型变化检测的正确率达到85%以上,对耕地、草地等植被类型也能取得较好的检测效果。     

多结构卷积神经网络特征级联的高分影像土地利用分类

高分辨率遥感影像包含丰富的土地利用类型信息,针对单一卷积神经网络提取图像特征信息不足的问题,提出了一种多结构卷积神经网络（convolutional neural network,CNN）特征级联的分类方法。首先,选择CaffeNet（convolutional architecture for fast feature embedding）、VGG-S（visual geometry group-slow）、VGG-F（visual geometry group-fast）为实验初始模型,对网络全连接层进行参数微调,采用随机梯度下降法（stochasticgradient descent,SGD）更新网络的权重;然后以微调后的网络分别作为特征提取器对图像提取特征,级联上述3种网络的第二个全连接层输出特征作为图像表达;最后,以多类最优边界分配机（multi-class optimal margindistribution machine,mcODM）获得最终分类结果。实验采用UC Merced land-use数据集进行分类效果检验,结果表明,多结构卷积神经网络级联的方法能够达到97.55%的总体分类精度,相较于CaffeNet、VGG-S和VGG-F等,分类精度分别提升了5.71%、2.72%和5.1%。因此多结构卷积神经网络特征级联的方法能够有效提取目标特征信息,提升土地利用分类精度。     

Change detection of surface mining activity and reclamation based on a machine learning approach of multi-temporal Landsat TM imagery

Being able to quantify land cover changes due to mining and reclamation at a watershed scale is of critical importance in managing and assessing their potential impacts to the Earth system. In this study, a remote sensing-based methodology is proposed for quantifying the impact of surface mining activity and reclamation from a watershed to local scale. The method is based on a Support Vector Machines (SVMs) classifier combined with multi-temporal change detection of Landsat TM imagery. The performance of the technique was evaluated at selected open mining sites located in the island of Milos in Greece. Assessment of the mining impact in the studied areas was based on the confusion matrix statistics, supported by co-orbital QuickBird-2 very high spatial resolution imagery. Overall classification accuracy of the thematic land cover maps produced was reported over 90%. Our analysis showed expansion of mining activity throughout the whole 23-year study period, while the transition of mining areas to soil and vegetation was evident in varying rates. Our results evidenced the ability of the method under investigation in deriving highly and accurate land cover change maps, able to identify the mining areas as well as those in which excavation was replaced by natural vegetation. All in all, the proposed technique showed considerable promise towards the support of a sustainable environmental development and prudent resource management.     

Modeling for Prediction of Land Cover Changes Based on Bio-physical and Human Factors in Zagros Mountains, Iran

The rapid population growth and ongoing development activities has resulted in natural resources demolition. However, the dynamics of the natural resources in relation to different biophysical and socio-economic factors are still remains poorly understood. The present study investigates the basic natural resources i.e. forest, rangeland and surface water bodies’ status using satellite data for the years 1990, 1998, and 2006, and their change detection in relation to biophysical and socio-economic factors. Monitoring land-use/cover change detection using remotely sensed data has been a well recognized technique. The analysis of change detection revealed eleven important land cover changes, which occurred during the past 16 years (1990–2006) in the region. The rate of land cover change was observed to vary across the sub periods and a general decline of forest cover and increase in rangelands and water bodies was observed. Logistic regression model was employed to analyze the relationship between changes and explanatory factors. The land cover change results and logistic models developed in this study are useful in supporting natural resources management efforts and provide useful information for managers/policy makers in formulation of sustainable management strategies for the region.     

地表覆盖遥感产品更新完善的研究动向

近年来,多尺度地表覆盖遥感产品的不断涌现,为环境变化研究、地球系统模拟、地理国(世)情监测和可持续发展规划等提供了重要科学数据。为更好地满足广大用户日益增长的应用需求,应对地表覆盖遥感产品进行持续更新完善,保持其时效性、增强时序性、丰富多样性。针对大面积地表覆盖遥感产品更新完善所面临的主要问题,介绍和评述了国内外有关研究动向,包括影像与众源信息相结合的更新、数据类型细化与完善、地表覆盖真实性验证,并作了简要展望。     

Multi-source change detection with PALSAR data in the Southern of Pará state in the Brazilian Amazon

Optical data is broadly used for change detection studies, despite being hindered by atmospheric conditions. Synthetic Aperture Radar (SAR) data can be useful for change detection in areas with frequent cloud coverage as SAR systems are capable of obtaining images almost independently from atmospheric conditions. This study aims to verify the difference in results of using SAR data instead of optical data for change detection purposes. Different levels of one hierarchical legend and both pixel and region-based classifiers were used. Change results were evaluated considering the use of rectangular matrices to incorporate the occurrence of impossible changes and relative comparison between change maps. Although the change maps obtained using only optical data were more accurate than those using either one or two land cover classifications based on L-band SAR data, the difference in the accuracy of change maps decreases with the use of less detailed legends. Additionally, results indicate that L-band SAR and multi-sensor approaches are adequate for deforestation identification even if post-classification results did not achieve global accuracy values superior to 0.86. The most accurate change detection results obtained in this work were not associated with the overall accuracy of land cover classifications, but with the distribution and accuracy of specific land cover classes.     

土地覆盖制图：基于最优化遥感数据的支撑向量机分类

遥感数据具有在不同空间、光谱和时间尺度上获取地表测量信息的能力,使其成为获取土地覆盖信息的一个主要数据源。影像分类即把卫星影像上的相关像元划分给某类已知的土地覆盖类型的过程。支撑向量机（SVMs）是一种土地覆盖分类的新技术。三种常用的SVMs是：基于线性和多项式的SVM以及具有高斯核函数的SVM分类器,分类能否成功地应用有赖于其各自选择的最佳参数。但是海量的遥感数据使得这些参数的确定速度十分缓慢。本文研究了一种新的基于最优化遥感数据压缩技术的SVM分类方法。研究显示用于获取SVM参数的数据量能够在不影响土地覆盖的分类精度的前提下进行压缩。数据压缩成功的应用于多项式和高斯核函数的SVM分类,而线性SVM的分类精度却非常低。     

Evaluation of different machine learning methods for land cover mapping of a Mediterranean area using multi-seasonal Landsat images and Digital Terrain Models

Land cover monitoring using digital Earth data requires robust classification methods that allow the accurate mapping of complex land cover categories. This paper discusses the crucial issues related to the application of different up-to-date machine learning classifiers: classification trees (CT), artificial neural networks (ANN), support vector machines (SVM) and random forest (RF). The analysis of the statistical significance of the differences between the performance of these algorithms, as well as sensitivity to data set size reduction and noise were also analysed. Landsat-5 Thematic Mapper data captured in European spring and summer were used with auxiliary variables derived from a digital terrain model to classify 14 different land cover categories in south Spain. Overall, statistically similar accuracies of over 91% were obtained for ANN, SVM and RF. However, the findings of this study show differences in the accuracy of the classifiers, being RF the most accurate classifier with a very simple parameterization. SVM, followed by RF, was the most robust classifier to noise and data reduction. Significant differences in their performances were only reached for thresholds of noise and data reduction greater than 20% (noise, SVM) and 25% (noise, RF), and 80% (reduction, SVM) and 50% (reduction, RF), respectively.     

顾及栅格空间拓扑关系的地表覆盖时空目标不一致性探测方法

地表覆盖与更新是地理国情监测、环境变化评估、生态系统保护等不可或缺的基础地理信息。遥感制图技术已成为地表覆盖信息提取的重要手段,但因地物光谱、纹理及时相等特征复杂性,地表覆盖更新数据往往存在错分、漏分,从而导致地表覆盖时空目标不一致。现有地表覆盖更新数据不一致性探测主要以人工检查为主、部分自动化为辅的方式,生产实践中需要大量的作业人员与时间,缺乏行之有效的不一致性自动化探测工具。本文研究分析了栅格地表覆盖更新数据不一致性检查面临的挑战,提出了基于复合逻辑量词的栅格空间拓扑关系计算方法、基于置信区间的更新期地表覆盖错分目标初判规则构建,以及利用空间约束多重匹配的更新期错分目标后验判断,形成了“关系-规则-判断”的地表覆盖时空目标不一致性探测体系。试验以山东临朐、垦利GlobeLand30数据为研究对象,经与统计一致性检核方法对比分析、参照真实地表影像数据,实现了地表覆盖时空目标不一致性探测与有效性检验,验证了探测方法可行性。     

遥感土地覆盖类型识别的自组织人工神经网络模型

本文提出了进行遥感土地覆盖类型识别的自组织人工神经网络模型,并选取了一组标准样作为研究对象,识别效率较高。结果表明,该网络性能良好,可望成为遥感土地覆盖类型识别的有效手段。     

Evaluating the Performance of Multi-Class and Single-Class Classification Approaches for Mountain Agriculture Extraction Using Time-Series NDVI

Supervised multi-class classification (MCC) approach is widely being used for regional-level land use–land cover (LULC) mapping and monitoring. However, it becomes inefficient if the end user wants to map only one particular class. Therefore, an improved single-class classification (SCC) approach is required for quick and reliable map production purpose. In this regard, the current study attempts to evaluate the performance of MCC and SCC approaches for extracting mountain agriculture area using time-series normalized differential vegetation index (NDVI). At first, samples of eight LULC classes were acquired using Google Earth image, and corresponding temporal signatures (TS) were extracted from time-series NDVI to perform classification using minimum distance to mean (MDM) and spectral angle mapper (i.e., multi-class SAM—MCSAM) under MCC approach. Secondly, under SCC approach, the TS of three agriculture classes (i.e., agriculture, mixed agriculture and plantation) were utilized as a reference to extract agriculture extent using Euclidean distance (ED) and SAM (i.e., single-class SAM—SCSAM) algorithms. The area of all four maps (i.e., MDM—19.77% of total geographical area (TGA), MCSAM—21.07% of TGA, ED—15.23% of TGA, SCSAM—13.85% of TGA) was compared with reference agriculture area (14.54% of TGA) of global land cover product, and SCC-based maps were found to have close agreement. Also, the class-wise detection accuracy was evaluated using random sample point-based error matrix which reveals the better performance of ED-based map than rest three maps in terms of overall accuracy and kappa coefficient.     

Assessing Land Cover Change in Hydro-Climatic Data Network Watersheds Using NALC Imagery

Land cover conversion is known to alter the hydrologic regimes of watersheds. While connections between land cover and runoff are generally known, not all land cover alterations result in detectable changes in streamflow, and the quantity of land cover change required to yield a detectable change in streamflow is unknown over a range of watersheds. The connection between land cover change and streamflow was explored for a Hydro-Climatic Data Network (HCDN) watershed. HCDN is a database of USGS gauged streams commonly used to assess the influence of climatic change on streamflow. Watersheds included in the HCDN have been screened to represent "unimpaired" streamflow. Implicit in this definition is the assumption that land cover is relatively unaltered over the streamflow time series. Imagery from the North American Landscape Characterization (NALC) project was analyzed to detect land cover change from 1972 to 1992 in an Oregon watershed selected from the HCDN. A post-classification change detection yielded a 44% rate of landscape change over 20 years. Changes in land cover classes by dominant soil types were paired with the L-THIA model of Purdue University to quantify the effect of land cover change on runoff. Despite land cover changes, simulations confirmed that runoff remained unchanged. This report summarizes recommended steps for applying NALC imagery to detection of landscape change in other watersheds.     

Evaluation of diverse classification approaches for land use/cover mapping in a Mediterranean region utilizing Hyperion data

Information on Earth's land surface cover is commonly obtained through digital image analysis of data acquired from remote sensing sensors. In this study, we evaluated the use of diverse classification techniques in discriminating land use/cover types in a typical Mediterranean setting using Hyperion imagery. For this purpose, the spectral angle mapper (SAM), the object-based and the non-linear spectral unmixing based on artificial neural networks (ANNs) techniques were applied. A further objective had been to investigate the effect of two approaches for training sites selection in the SAM classification, namely of the pixel purity index (PPI) and of the direct selection of training points from the Hyperion imagery assisted by a QuickBird imagery and field-based training sites. Object-based classification outperformed the other techniques with an overall accuracy of 83%. Sub-pixel classification based on the ANN showed an overall accuracy of 52%, very close to that of SAM (48%). SAM applied using the training sites selected directly from the Hyperion imagery supported by the QuickBird image and the field visits returned an increase accuracy by 16%. Yet, all techniques appeared to suffer from the relatively low spatial resolution of the Hyperion imagery, which affected the spectral separation among the land use/cover classes.     

引入二维交细分类型的地表覆盖矢量数据增量更新

通过遥感影像与基准年数据对比获得变化信息是目前地表覆盖数据增量更新的主要变化数据来源,但现有方法不能直接更新地表覆盖矢量数据。本文设计了一种包含变化对象的空间位置和类型信息的地表覆盖增量数据模型,发展了一种引入面/面二维交细分类型的地表覆盖矢量数据增量更新方法。该方法首先采用基于目标整体交、差结果的欧拉数的E-WID层次拓扑关系模型区,分析了地表覆盖矢量数据更新中的14种二维交细分拓扑关系类型;然后根据这些二维交细分类型,设计了9条自动更新处理规则。最后开发了一套基于根据二维交细分类型处理规则的地表覆盖数据增量更新原型系统,并用实际数据验证了其正确性。     

基于神经网络趋势面分析的地价样点检验方法研究

建立了基于人工神经网络的地价趋势面分析模型,提出了基于该模型并结合可视化检视进行地价样点粗差检测的算法,实例验证了该方法的可行性.     

Integrating RapidEye and ancillary data to map alpine habitats in South Tyrol,Italy

In this paper, we present a two-stage method for mapping habitats using Earth observation (EO) data in three Alpine sites in South Tyrol, Italy. The first stage of the method was the classification of land cover types using multi-temporal RapidEye images and support vector machines (SVMs). The second stage involved reclassification of the land cover types to habitat types following a rule-based spatial kernel. The highest accuracies in land cover classification were 95.1% overall accuracy, 0.94 kappa coefficient and 4.9% overall disagreement. These accuracies were obtained when the combination of images with topographic parameters and homogeneity texture was used. The habitat classification accuracies were rather moderate due to the broadly defined rules and possible inaccuracies in the reference map. Overall, our proposed methodology could be implemented to map cost-effectively alpine habitats over large areas and could be easily adapted to map other types of habitats.