Unsupervised classification of remote sensing images combining Self Organizing Maps and segmentation techniques

Dissertation submitted in partial fulfillment of the requirements for the Degree of Master of Science in Geospatial Technologies.

Structured output SVM for remote sensing image classification

In the recent years, kernel methods have revealed very powerful tools in many application domains in general and in remote sensing image classification in particular. The special characteristics of remote sensing images (high dimension, few labeled samples and different noise sources) are efficiently dealt with kernel machines. In this paper, we propose the use of structured output learning to improve remote sensing image classification based on kernels. Structured output learning is concerned with the design of machine learning algorithms that not only implement input-output mapping, but also take into account the relations between output labels, thus generalizing unstructured kernel methods. We analyze the framework and introduce it to the remote sensing community. Output similarity is here encoded into SVM classifiers by modifying the model loss function and the kernel function either independently or jointly. Experiments on a very high resolution (VHR) image classification problem shows promising results and opens a wide field of research with structured output kernel methods.

Web-based Implementation of Winter Maintenance Decision Support System Using GIS and Remote Sensing, May 2005

Winter maintenance, particularly snow removal and the stress of snow removal materials on public structures, is an enormous budgetary burden on municipalities and nongovernmental maintenance organizations in cold climates. Lately, geospatial technologies such as remote sensing, geographic information systems (GIS), and decision support tools are roviding a valuable tool for planning snow removal operations. A few researchers recently used geospatial technologies to develop winter maintenance tools. However, most of these winter maintenance tools, while having the potential to address some of these information needs, are not typically placed in the hands of planners and other interested stakeholders. Most tools are not constructed with a nontechnical user in mind and lack an easyto-use, easily understood interface. A major goal of this project was to implement a web-based Winter Maintenance Decision Support System (WMDSS) that enhances the capacity of stakeholders (city/county planners, resource managers, transportation personnel, citizens, and policy makers) to evaluate different procedures for managing snow removal assets optimally. This was accomplished by integrating geospatial analytical techniques (GIS and remote sensing), the existing snow removal asset management system, and webbased spatial decision support systems. The web-based system was implemented using the ESRI ArcIMS ActiveX Connector and related web technologies, such as Active Server Pages, JavaScript, HTML, and XML. The expert knowledge on snow removal procedures is gathered and integrated into the system in the form of encoded business rules using Visual Rule Studio. The system developed not only manages the resources but also provides expert advice to assist complex decision making, such as routing, optimal resource allocation, and monitoring live weather information. This system was developed in collaboration with Black Hawk County, IA, the city of Columbia, MO, and the Iowa Department of transportation. This product was also demonstrated for these agencies to improve the usability and applicability of the system.

Active learning methods for remote sensing image classification

In this paper, we propose two active learning algorithms for semiautomatic definition of training samples in remote sensing image classification. Based on predefined heuristics, the classifier ranks the unlabeled pixels and automatically chooses those that are considered the most valuable for its improvement. Once the pixels have been selected, the analyst labels them manually and the process is iterated. Starting with a small and nonoptimal training set, the model itself builds the optimal set of samples which minimizes the classification error. We have applied the proposed algorithms to a variety of remote sensing data, including very high resolution and hyperspectral images, using support vector machines. Experimental results confirm the consistency of the methods. The required number of training samples can be reduced to 10% using the methods proposed, reaching the same level of accuracy as larger data sets. A comparison with a state-of-the-art active learning method, margin sampling, is provided, highlighting advantages of the methods proposed. The effect of spatial resolution and separability of the classes on the quality of the selection of pixels is also discussed.

Domain Adaptation in remote sensing: increasing the portability of land-cover classifiers

Among the types of remote sensing acquisitions, optical images are certainly one of the most widely relied upon data sources for Earth observation. They provide detailed measurements of the electromagnetic radiation reflected or emitted by each pixel in the scene. Through a process termed supervised land-cover classification, this allows to automatically yet accurately distinguish objects at the surface of our planet. In this respect, when producing a land-cover map of the surveyed area, the availability of training examples representative of each thematic class is crucial for the success of the classification procedure. However, in real applications, due to several constraints on the sample collection process, labeled pixels are usually scarce. When analyzing an image for which those key samples are unavailable, a viable solution consists in resorting to the ground truth data of other previously acquired images. This option is attractive but several factors such as atmospheric, ground and acquisition conditions can cause radiometric differences between the images, hindering therefore the transfer of knowledge from one image to another. The goal of this Thesis is to supply remote sensing image analysts with suitable processing techniques to ensure a robust portability of the classification models across different images. The ultimate purpose is to map the land-cover classes over large spatial and temporal extents with minimal ground information. To overcome, or simply quantify, the observed shifts in the statistical distribution of the spectra of the materials, we study four approaches issued from the field of machine learning. First, we propose a strategy to intelligently sample the image of interest to collect the labels only in correspondence of the most useful pixels. This iterative routine is based on a constant evaluation of the pertinence to the new image of the initial training data actually belonging to a different image. Second, an approach to reduce the radiometric differences among the images by projecting the respective pixels in a common new data space is presented. We analyze a kernel-based feature extraction framework suited for such problems, showing that, after this relative normalization, the cross-image generalization abilities of a classifier are highly increased. Third, we test a new data-driven measure of distance between probability distributions to assess the distortions caused by differences in the acquisition geometry affecting series of multi-angle images. Also, we gauge the portability of classification models through the sequences. In both exercises, the efficacy of classic physically- and statistically-based normalization methods is discussed. Finally, we explore a new family of approaches based on sparse representations of the samples to reciprocally convert the data space of two images. The projection function bridging the images allows a synthesis of new pixels with more similar characteristics ultimately facilitating the land-cover mapping across images.

Advanced Kernel Methods For Remote Sensing Image Classification

Résumé Suite aux recentes avancées technologiques, les archives d'images digitales ont connu une croissance qualitative et quantitative sans précédent. Malgré les énormes possibilités qu'elles offrent, ces avancées posent de nouvelles questions quant au traitement des masses de données saisies. Cette question est à la base de cette Thèse: les problèmes de traitement d'information digitale à très haute résolution spatiale et/ou spectrale y sont considérés en recourant à des approches d'apprentissage statistique, les méthodes à noyau. Cette Thèse étudie des problèmes de classification d'images, c'est à dire de catégorisation de pixels en un nombre réduit de classes refletant les propriétés spectrales et contextuelles des objets qu'elles représentent. L'accent est mis sur l'efficience des algorithmes, ainsi que sur leur simplicité, de manière à augmenter leur potentiel d'implementation pour les utilisateurs. De plus, le défi de cette Thèse est de rester proche des problèmes concrets des utilisateurs d'images satellite sans pour autant perdre de vue l'intéret des méthodes proposées pour le milieu du machine learning dont elles sont issues. En ce sens, ce travail joue la carte de la transdisciplinarité en maintenant un lien fort entre les deux sciences dans tous les développements proposés. Quatre modèles sont proposés: le premier répond au problème de la haute dimensionalité et de la redondance des données par un modèle optimisant les performances en classification en s'adaptant aux particularités de l'image. Ceci est rendu possible par un système de ranking des variables (les bandes) qui est optimisé en même temps que le modèle de base: ce faisant, seules les variables importantes pour résoudre le problème sont utilisées par le classifieur. Le manque d'information étiquétée et l'incertitude quant à sa pertinence pour le problème sont à la source des deux modèles suivants, basés respectivement sur l'apprentissage actif et les méthodes semi-supervisées: le premier permet d'améliorer la qualité d'un ensemble d'entraînement par interaction directe entre l'utilisateur et la machine, alors que le deuxième utilise les pixels non étiquetés pour améliorer la description des données disponibles et la robustesse du modèle. Enfin, le dernier modèle proposé considère la question plus théorique de la structure entre les outputs: l'intègration de cette source d'information, jusqu'à présent jamais considérée en télédétection, ouvre des nouveaux défis de recherche. Advanced kernel methods for remote sensing image classification Devis Tuia Institut de Géomatique et d'Analyse du Risque September 2009 Abstract The technical developments in recent years have brought the quantity and quality of digital information to an unprecedented level, as enormous archives of satellite images are available to the users. However, even if these advances open more and more possibilities in the use of digital imagery, they also rise several problems of storage and treatment. The latter is considered in this Thesis: the processing of very high spatial and spectral resolution images is treated with approaches based on data-driven algorithms relying on kernel methods. In particular, the problem of image classification, i.e. the categorization of the image's pixels into a reduced number of classes reflecting spectral and contextual properties, is studied through the different models presented. The accent is put on algorithmic efficiency and the simplicity of the approaches proposed, to avoid too complex models that would not be used by users. The major challenge of the Thesis is to remain close to concrete remote sensing problems, without losing the methodological interest from the machine learning viewpoint: in this sense, this work aims at building a bridge between the machine learning and remote sensing communities and all the models proposed have been developed keeping in mind the need for such a synergy. Four models are proposed: first, an adaptive model learning the relevant image features has been proposed to solve the problem of high dimensionality and collinearity of the image features. This model provides automatically an accurate classifier and a ranking of the relevance of the single features. The scarcity and unreliability of labeled. information were the common root of the second and third models proposed: when confronted to such problems, the user can either construct the labeled set iteratively by direct interaction with the machine or use the unlabeled data to increase robustness and quality of the description of data. Both solutions have been explored resulting into two methodological contributions, based respectively on active learning and semisupervised learning. Finally, the more theoretical issue of structured outputs has been considered in the last model, which, by integrating outputs similarity into a model, opens new challenges and opportunities for remote sensing image processing.

Multispectral remote sensing for site-specific nitrogen fertilizer management

The objective of this work was to evaluate the use of multispectral remote sensing for site-specific nitrogen fertilizer management. Satellite imagery from the advanced spaceborne thermal emission and reflection radiometer (Aster) was acquired in a 23 ha corn-planted area in Iran. For the collection of field samples, a total of 53 pixels were selected by systematic randomized sampling. The total nitrogen content in corn leaf tissues in these pixels was evaluated. To predict corn canopy nitrogen content, different vegetation indices, such as normalized difference vegetation index (NDVI), soil-adjusted vegetation index (Savi), optimized soil-adjusted vegetation index (Osavi), modified chlorophyll absorption ratio index 2 (MCARI2), and modified triangle vegetation index 2 (MTVI2), were investigated. The supervised classification technique using the spectral angle mapper classifier (SAM) was performed to generate a nitrogen fertilization map. The MTVI2 presented the highest correlation (R²=0.87) and is a good predictor of corn canopy nitrogen content in the V13 stage, at 60 days after cultivating. Aster imagery can be used to predict nitrogen status in corn canopy. Classification results indicate three levels of required nitrogen per pixel: low (0-2.5 kg), medium (2.5-3 kg), and high (3-3.3 kg).

Using GPS, GIS & remote sensing to understand Niagara Terroir : Pinot noir in the Four Mile Creek & St. David's Bench sub-appellations

The relationships between vine water status, soil texture, and vine size were observed in four Niagara, Ontario Pinot noir vineyards in 2008 and 2009. The vineyards were divided into water status zones using geographic information systems (GIS) software to map the seasonal mean midday leaf water potential (,P), and dormant pruning shoot weights following the 2008 season. Fruit was harvested from all sentinel vines, bulked by water status zones and made into wine. Sensory analysis included a multidimensional sorting (MDS) task and descriptive analysis (DA) of the 2008 wines. Airborne multispectral images, with a spatial resolution of 38 cm, were captured four times in 2008 and three times in 2009, with the final flights around veraison. A semi-automatic process was developed to extract NDVI from the images, and a masking procedure was identified to create a vine-only NDVI image. 2008 and 2009 were cooler and wetter than mean years, and the range of water status zones was narrow. Yield per vine, vine size, anthocyanins and phenols were the least consistent variables. Divided by water status or vine size, there were no variables with differences between zones in all four vineyards in either year. Wines were not different between water status zones in any chemical analysis, and HPLC revealed that there were no differences in individual anthocyanins or phenolic compounds between water status zones within the vineyard sites. There were some notable correlations between vineyard and grape composition variables, and spatial trends were observed to be qualitatively related for many of the variables. The MDS task revealed that wines from each vineyard were more affected by random fermentation effects than water status effects. This was confirmed by the DA; there were no differences between wines from the water status zones within vineyard sites for any attribute. Remotely sensed NDVI (normalized difference vegetation index) correlated reasonably well with a number of grape composition variables, as well as soil type. Resampling to a lower spatial resolution did not appreciably affect the strength of correlations, and corresponded to the information contained in the masked images, while maintaining the range of values of NDVI. This study showed that in cool climates, there is the potential for using precision viticulture techniques to understand the variability in vineyards, but the variable weather presents a challenge for understanding the driving forces of that variability.

Using GPS, GIS & remote sensing to understand Niagara Terroir

The focus of this study was to detennine whether soil texture and composition variables were related to vine water status and both yield components and grape composition, and whether multispectral high definition airborne imagery could be used to segregate zones in vineyards to target fruit of highest quality for premium winemaking. The study took place on a 10-ha commercial Riesling vineyard at Thirty Bench Winemakers, in Beamsville (Ontario). Results showed that Soil moisture and leaf'l' were temporally stable and related to berry composition and remotely-sensed data. Remote-sensing, through the calculation of vegetation indices, was particularly useful to predict vine vigor, yield, fruit maturity as well as berry monoterpene concentration; it could also clearly assist in making wines that are more representative ofthe cultivar used, and also wines that are a reflection of a specific terroir, since calculated vegetation indices were highly correlated to typical Riesling.