Exploring uncertainties in terrain feature extraction across multi-scale,multi-feature,and multi-method approaches for variable terrain

Terrain analysis uses different workflows to extract features from terrain models for the purpose of understanding topographic patterns and processes. However, the results of different workflows often conflict, leading to uncertainties about feature locations. Instead of relying upon a single workflow, we suggest that a fusion of information from multiple workflows better informs terrain analysis. From terrain data with different degrees of variability, we extracted terrain features related to the set of topographic surface network feature classes {peaks, pits, saddles, ridges, courses} using workflows from free, open-source, and commercial software. A multi-scale analysis produced terrain features with fuzzy membership values for various feature classes and revealed that terrain locations can exhibit characteristics of all classes. Multi-feature maps were created by determining at each location the dominant and second-ranked features, and an uncertainty value. Our multi-method approach incorporated all of the workflows’ multi-scale results and again produced multi-feature maps that increased the confidence of some features and reduced the signal of dissimilar results. We also found that high variability terrain produced crisper features in both spatial extent and membership strength. Our overall conclusion is that multi-scale, multi-feature, and multi-method analyses clarify terrain feature uncertainty.     

Object-Based Land Cover Classification Using High-Posting-Density LiDAR Data

This study introduces a method for object-based land cover classification based solely on the analysis of LiDAR-derived information—i.e., without the use of conventional optical imagery such as aerial photography or multispectral imagery. The method focuses on the relative information content from height, intensity, and shape of features found in the scene. Eight object-based metrics were used to classify the terrain into land cover information: mean height, standard deviation (STDEV) of height, height homogeneity, height contrast, height entropy, height correlation, mean intensity, and compactness. Using machine-learning decision trees, these metrics yielded land cover classification accuracies > 90%. A sensitivity analysis found that mean intensity was the key metric for differentiating between the grass and road/parking lot classes. Mean height was also a contributing discriminator for distinguishing features with different height information, such as between the building and grass classes. The shape- or texture-based metrics did not significantly improve the land cover classifications. The most important three metrics (i.e., mean height, STDEV height, and mean intensity) were sufficient to achieve classification accuracies > 90%.     

Forward modeling of GPS multipath for near-surface reflectometry and positioning applications

Multipath is detrimental for both GPS positioning and timing applications. However, the benefits of GPS multipath for reflectometry have become increasingly clear for soil moisture, snow depth, and vegetation growth monitoring. Most multipath forward models focus on the code modulation, adopting arbitrary values for the reflection power, phase, and delay, or they calculate the reflection delay based on a given geometry and keep reflection power empirically defined. Here, a fully polarimetric forward model is presented, accounting for right- and left-handed circularly polarized components of the GPS broadcast signal and of the antenna and surface responses as well. Starting from the fundamental direct and reflected voltages, we have defined the interferometric and error voltages, which are of more interest in reflectometry and positioning applications. We examined the effect of varying coherence on signal-to-noise ratio, carrier phase, and code pseudorange observables. The main features of the forward model are subsequently illustrated as they relate to the broadcast signal, reflector height, random surface roughness, surface material, antenna pattern, and antenna orientation. We demonstrated how the antenna orientation—upright, tipped, or upside-down—involves a number of trade-offs regarding the neglect of the antenna gain pattern, the minimization of CDMA self-interference, and the maximization of the number of satellites visible. The forward model was also used to understand the multipath signature in GPS positioning applications. For example, we have shown how geodetic GPS antennas offer little impediment for the intake of near-grazing reflections off natural surfaces, in contrast to off metal, because of the lack of diversity with respect to the direct signal—small interferometric delay and Doppler, same sense of polarization, and similar direction of arrival.     

Quality Assessment Of Digital Surface Models Generated From IKONOS Imagery

The growing applications of digital surface models (DSMs) for object detection, segmentation and representation of terrestrial landscapes have provided impetus for further automation of 3D spatial information extraction processes. While new technologies such as lidar are available for almost instant DSM generation, the use of stereoscopic high-resolution satellite imagery (HRSI), coupled with image matching, affords cost-effective measurement of surface topography over large coverage areas. This investigation explores the potential of IKONOS Geo stereo imagery for producing DSMs using an alternative sensor orientation model, namely bias-corrected rational polynomial coefficients (RPCs), and a hybrid image-matching algorithm. To serve both as a reference surface and a basis for comparison, a lidar DSM was employed in the Hobart testfield, a region of differing terrain types and slope. In order to take topographic variation within the modelled surface into account, the lidar strip was divided into separate sub-areas representing differing land cover types. It is shown that over topographically diverse areas, heighting accuracy to better than 3 pixels can be readily achieved. Results improve markedly in feature-rich open and relatively flat terrain, with sub-pixel accuracy being achieved at check points surveyed using the global positioning system (GPS). This assessment demonstrates that the outlook for DSM generation from HRSI is very promising.     

Texture-based classification of sub-Antarctic vegetation communities on Heard Island

This study was the first to use high-resolution IKONOS imagery to classify vegetation communities on sub-Antarctic Heard Island. We focused on the use of texture measures, in addition to standard multispectral information, to improve the classification of sub-Antarctic vegetation communities. Heard Island’s pristine and rapidly changing environment makes it a relevant and exciting location to study the regional effects of climate change. This study uses IKONOS imagery to provide automated, up-to-date, and non-invasive means to map vegetation as an important indicator for environmental change. Three classification techniques were compared: multispectral classification, texture based classification, and a combination of both. Texture features were calculated using the Grey Level Co-occurrence Matrix (GLCM). We investigated the effect of the texture window size on classification accuracy. The combined approach produced a higher accuracy than using multispectral bands alone. It was also found that the selection of GLCM texture features is critical. The highest accuracy (85%) was produced using all original spectral bands and three uncorrelated texture features. Incorporating texture improved classification accuracy by 6%.     

地形引起的雷达辐射畸变及其校正

合成孔径雷达影像由于其侧视特点 ,存在着严重的地形引起的几何畸变及辐射畸变。辐射畸变不仅对 SAR辐射标定造成困难 ,而且严重影响了影像分类、土壤湿度信息提取、森林蓄积量信息提取等应用。本文将辐射畸变归结为面积效应和局部入射角效应 ,推导了散射面积归一化因子 ,以消除辐射畸变的面积效应。提出了一种以局部入射角的线性函数表达的后向散射模型 ,在此基础上 ,给出了消除局部入射角效应的校正函数。最后 ,以RADARSAT SAR影像进行地形辐射畸变校正的试验与分析     

Cloud classification in NOAA AVHRR imageries using spectral and textural features

Clouds contribute significantly to the formation of many of the natural hazards. Hence cloud mapping and its classification becomes a major component of the various physical models which are used for forecasting natural hazards. The problem of cloud data classification from NOAA AVHRR (advance very high resolution radiometer) satellite imagery using image transformation techniques is considered in this paper. The singular value decomposition (SVD) scheme is used to extract the salient spectral and textural features attributed to satellite snow and cloud data in visible and IR channels. The goals of this paper are to discriminate between clear sky and clouds in an 8 × 8 pixel array of 1.1 km AVHRR data. If clouds are present then classify them as low, medium or high range. This scheme can effectively segregate clouds and non-cloud features in the visible and IR bands of the imagery. It can also classify clouds as low, medium or high range with a success rate of 70–90%. Computer-based snow and cloud discrimination and automatic cloud classification system will help the forecaster in various climatological applications, viz., energy balance estimation, precipitation forecasting, landslide forecasting, weather forecasting and avalanche forecasting etc.     

多层级二值模式的高光谱影像空-谱分类

利用高光谱遥感影像的空间纹理特征,可以提高高光谱遥感影像的分类精度。提出了一种多层级二值模式的高光谱影像空-谱联合分类方法。该方法将高光谱影像转化为局部二值模式特征图像获取像元微观特征,基于特征图像生成多层级特征向量获取像元宏观特征。为验证该方法的有效性,选取PaviaU、Salinas和Chikusei高光谱影像数据,利用核极限学习机分类器,分别针对光谱、局部二值模式、多层级二值模式等特征开展实验。结果表明,多层级二值模式空-谱分类总体精度分别达到97.31%、98.96%和97.85%,明显优于传统光谱、3Gabor空-谱等分类方法。该方法可为高光谱影像分类提供更加有效的类别判定特征,有助于提高影像分类精度并获取更加平滑的分类结果图。     

Class imbalance in unsupervised change detection – A diagnostic analysis from urban remote sensing

Automatic monitoring of changes on the Earth’s surface is an intrinsic capability and simultaneously a persistent methodological challenge in remote sensing, especially regarding imagery with very-high spatial resolution (VHR) and complex urban environments. In order to enable a high level of automatization, the change detection problem is solved in an unsupervised way to alleviate efforts associated with collection of properly encoded prior knowledge. In this context, this paper systematically investigates the nature and effects of class distribution and class imbalance in an unsupervised binary change detection application based on VHR imagery over urban areas. For this purpose, a diagnostic framework for sensitivity analysis of a large range of possible degrees of class imbalance is presented, which is of particular importance with respect to unsupervised approaches where the content of images and thus the occurrence and the distribution of classes are generally unknown a priori. Furthermore, this framework can serve as a general technique to evaluate model transferability in any two-class classification problem. The applied change detection approach is based on object-based difference features calculated from VHR imagery and subsequent unsupervised two-class clustering using k‐means, genetic k-means and self-organizing map (SOM) clustering. The results from two test sites with different structural characteristics of the built environment demonstrated that classification performance is generally worse in imbalanced class distribution settings while best results were reached in balanced or close to balanced situations. Regarding suitable accuracy measures for evaluating model performance in imbalanced settings, this study revealed that the Kappa statistics show significant response to class distribution while the true skill statistic was widely insensitive to imbalanced classes. In general, the genetic k-means clustering algorithm achieved the most robust results with respect to class imbalance while the SOM clustering exhibited a distinct optimization towards a balanced distribution of classes.     

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.     

Using Thermal Remote Sensing in Environmental Studies

Thermovision is a relatively new method of remote sensing with applications in areas such as military operations, residential monitoring, technological process control and emergency management. Surprisingly, it has not seen much application in environmental studies. The article presents a method of using thermovision for topoclimatic studies. The method is based on the spatial distribution of land surface temperature (LST). The LST distribution indicates the amount of solar energy reaching the Earth surface and depends primarily on terrain shape and land cover types. By analyzing the LST distribution, one can determine spatial topoclimatic variability. The LST derived topoclimatic classification was compared with the theoretical topoclimatic classification based on heat balance. New classes of topoclimates were created and some of the existing types were diversified into more detailed subtypes. The analysis of selected lowland areas in north‐western Poland revealed that both land cover and terrain shape characteristics had a significant impact on the LST distribution, contrary to the expectation of land cover characteristics being more important than terrain shape. The article demonstrates the possibilities of using thermovision in environmental research and presents a new method of topoclimate delimitation based on thermal remote sensing data and geographical information systems (GIS) techniques comparing. The LST classification method with conventional methods based on DEM and land cover analysis.     

Visual interpretation of landsat-1 imagery of central and southern districts of Haryana state,India

LANDSAT-MSS data in the form of a false colour composite image at a scale of 1:250,000 has been used to produce terrain unit maps of parts of north west India. The area of study consisted of central and southern districts of Haryana State. It has been possible to obtain a qualitative assessment of land use patterns and surface hydrology by means of visual Interpretation. The boundaries between differeat terrain units and moisture regimes were clearly discernible It is felt that comparative studies of false colour infrared imagery over a period of time can provide valuable Information for those engaged in reclamation schemes. In particular, the data can be used to plan reclamation of salt affected and waterlogged lands in semi-arid zones of states like Haryana.     

机载GPS反射信号土壤湿度测量技术

随着全球导航定位系统反射信号(GNSS-R)技术的发展, 近年来提出了利用GPS地表反射信号遥感土壤湿度的新方法, 该方法利用地表反射率与土壤介电常数以及介电常数与土壤湿度之间的关系来建立反演模型。为了可以快速方便的利用DMR实测数据反演得到土壤湿度, 本文根据Wang和Schmugge模型建立了土壤介电常数与湿度之间的分段模型, 实现了从原始反射数据到土壤湿度结果的整个反演流程。为了验证反演的可行性, 利用NASA等机构联合进行的SMEX02试验机载数据反演得到的结果表明, GPS反射信号能够有效地反演     

Age-based classification of arecanut crops: a case study of Channagiri,Karnataka, India

Arecanut is one of the predominant plantation crop grown in India. Yield of this crop depends upon age of the crop and there is no information on the spectral behaviour of arecanut crops across its ages. In this study popular supervised classification algorithms were utilized for age discrimination of arecanut crops using Hyperion imagery. Arecanut plantations selected for the study are located in Channagiri Taluk, Davanagere district of Karnataka state, India. Ground truth information collected involves: (i) GPS coordinates of selected plots, (ii) spectral reflectance of arecanut crops with age ranging from 1 to 50 years, using handheld spectroradiometer with 1 nm spectral resolution. These spectral measurements were made close in time to the acquisition of Hyperion imagery to build age-based spectral library. It is observed from the analysis that crops of ages below 3, 3–7, 8–15 and above 15 years were showing distinct spectral behaviour. Accordingly, crops age ranging from 1 to 50 were grouped into four classes. Classification of arecanut crops based on age groups was performed using methods like spectral angle mapper, support vector machine and minimum distance classifier, and were compared to find the most suitable method. Among the classification methods adopted, support vector machine with linear kernel function resulted in most accurate classification method with overall accuracy of 72% for within class seperability. Individual age group classification producer’s accuracy varied minimum of 12.5% for 3–7 years age group and maximum of 86.25% for above 15 years age group. It may be concluded that, not only age- based arecanut crop classification is possible, but also it is possible to develop age-based spectral library for plantation crop like arecanut.     

Building population mapping with aerial imagery and GIS data

Geospatial distribution of population at a scale of individual buildings is needed for analysis of people's interaction with their local socio-economic and physical environments. High resolution aerial images are capable of capturing urban complexities and considered as a potential source for mapping urban features at this fine scale. This paper studies population mapping for individual buildings by using aerial imagery and other geographic data. Building footprints and heights are first determined from aerial images, digital terrain and surface models. City zoning maps allow the classification of the buildings as residential and non-residential. The use of additional ancillary geographic data further filters residential utility buildings out of the residential area and identifies houses and apartments. In the final step, census block population, which is publicly available from the U.S. Census, is disaggregated and mapped to individual residential buildings. This paper proposes a modified building population mapping model that takes into account the effects of different types of residential buildings. Detailed steps are described that lead to the identification of residential buildings from imagery and other GIS data layers. Estimated building populations are evaluated per census block with reference to the known census records. This paper presents and evaluates the results of building population mapping in areas of West Lafayette, Lafayette, and Wea Township, all in the state of Indiana, USA.     

Combination of Spectral and Textural Features in the MSG Satellite Remote Sensing Images for Classifying Rainy Area into Different Classes

The rainfall intensity classification technique using spectral and textural features from MSG/SEVIRI (Meteosat Second Generation/Spinning Enhanced Visible and Infrared) satellite data is proposed in this paper. The study is carried out over north of Algeria. The developed method is based on the artificial neural multilayer perceptron network (MLP). Two MLP algorithms are used: the MLP-S based only on spectral parameters and the MLP-ST that use both spectral and textural features. The MLP model is created with three layers (input, hidden, and output) that consist of 6 output neurons in the output layer that represent the 6 rain intensities classes: very high, moderate to high, moderate, light to moderate, light and no rain and 10 spectral input neurons for the MLP-S and 15 input neurons for MLP-ST, which as ten spectral features that were calculated from MSG thermal infrared brilliance temperature and brilliance temperature difference and as five textural features, and The rainfall intensity areas classified by the proposed technique are validated against ground-based radar data. The rainfall rates used in the training set are derived from Setif radar measurements (Algeria). The results obtained after applying this method show that the introduction of textural parameters as additional information works in improving the classification of different rainfall intensities pixels in the MSG/SEVIRI imagery compared to the techniques based only on spectral information. These results are compared with results obtained with the probability of rainfall intensity (PRI). This comparison revealed a clear outperformance of the MLP algorithms over the PRI algorithms. Best results are provided by the MLP-ST algorithm. The combination of spectral and textural features in the MSG–SEVIRI imagery is important and for the classification of the rainfall intensities to different classes.     

遥感图像土地覆盖分类中多源特征数据选择研究

多源特征数据可以提高遥感图像的分类精度,选择合适的特征数据十分重要。利用基尼指数对多尺度纹理信息、主成分变换前三分量、地形数据等特征进行选择,选出最佳特征子集。利用支持向量机、神经网络分类法、最大似然法分别对全部特征数据和最佳特征子集结合多光谱数据进行分类。实验结果表明:基尼指数可以有效地对多源特征数据进行选择,特征选择可以提高分类器效率,提高分类精度。     

高光谱影像的DAE分类

针对高光谱影像非线性分类问题,根据高光谱影像光谱分辨率高且光谱具有非线性的特点,结合深度学习理论,提出了一种采用降噪自动编码器(DAE)的高光谱影像分类方法。该方法结合降噪自动编码器与SOFTMAX分类器,构造深层网络分类模型;然后,利用加噪后的光谱数据,采用Dropout方法对分类模型进行预训练和微调;最后,利用训练得到的网络模型学习高光谱影像光谱的隐含特征,实现高光谱影像的分类。采用该方法对AVIRIS和PHI的高光谱影像分别进行分类对比实验,结果表明该方法能有效提高高光谱影像分类精度。     

Object-based classification with features extracted by a semi-automatic feature extraction algorithm – SEaTH

Object-based image analysis (OBIA) uses object features (or attributes) that relateto the pixels contained by the image object to assist in image classification. These object features include spectral, shape, texture and context features. With hundreds of available features, the identification of those that can improve separability between classes is critical for OBIA. The Separability and Thresholds (SEaTH) algorithm calculates the SEaTH of object–classes for the given features. The SEaTH algorithm avoids time-consuming trial-and-error practice for seeking important features and thresholds. This article tests the SEaTH algorithm on Landsat-7 Enhanced Thematic Mapper (ETM+) imagery in a heterogeneous landscape with multiple land cover classes. The results suggest SEaTH is a strong alternative to other automated approaches, yielding an agreement of 79% with reference data. In comparison, an object-based nearest neighbour classifier yielded 66% agreement and a pixel-based maximum likelihood classifier yielded 69% agreement.     

Glacier surface characteristics derivation and monitoring using Hyperspectral datasets: a case study of Gepang Gath glacier,Western Himalaya

Mountain Glaciers are natural resources of fresh water and these affect the stream flow of the rivers, regional climate and further global climate. Observed trends and projected future evolutions of climate and Cryospheric variables clearly suggest a need to monitor these changes. Accordingly, the article presents the glacier features mapping using Hyperspectral remote sensing imagery. A freely available Hyperion satellite imagery acquired over Gepang Gath glacier in Himachal Pradesh, India is used for the study. Each class is identified based on their surface characteristics of spectral reflectance properties. Identification is simplified by demarcating the study glacier into accumulation and ablation areas through snowline. Accumulation area is characterized with high reflectance clean snow/ice and reduced moderate reflectance Snow/firn. The identification of classes in Hyperion imagery is validated using the spectral library from USGS and ASTER, and field spectra obtained from literature.