A análise de cenas complexas por computadores não é uma tarefa trivial, entretanto, o cérebro humano pode realizar esta função de maneira eficiente. A evolução natural tem desenvolvido formas para otimizar nosso sistema visual de modo que apenas partes importantes da cena sejam analisadas a cada instante. Este mecanismo de seleção é denominado por atenção visual. A atenção visual opera sob dois aspectos: bottom-up e top-down. A atenção bottom-up é dirigida por conspicuidades baseadas na cena, como o contraste de cores, orientação, etc. Por outro lado, a atenção top-down é controlada por tarefas, memórias, etc. A atenção top-down pode ainda modular o mecanismo bottom-up através do enviesamento de determinadas características de acordo com a tarefa. Além do mecanismo de modulação considerado, o que é selecionado a partir da cena também representa uma importante parte para o processo de seleção. Neste cenário, diversas teorias têm sido propostas e podem ser agrupadas em duas linhas principais: atenção baseada no espaço e atenção baseada em objetos. Modelos baseados em objeto, ao invés de apenas direcionar a atenção para locais ou características específicas da cena, requerem que a seleção seja realizada a nível de objeto, significando que os objetos são a unidade básica da percepção. De modo a desenvolver modelos de acordo com a teoria baseada em objetos, deve-se considerar a integração de um módulo de organização perceptual. Este módulo pode segmentar os objetos do fundo da cena baseado em princípios de agrupamento tais como similaridade, proximidade, etc. Esses objetos competirão pela atenção. Diversos modelos de atenção visual baseados em objetos tem sido propostos nos últimos anos. Pesquisas em modelos de atenção visual têm sido desenvolvidas principalmente relacionadas à atenção bottom-up guiadas por características visuais primitivas, desconsiderando qualquer informação sobre os objetos. Por outro lado, trabalhos recentes têm sido realizados em relação ao uso do conhecimento sobre o alvo para influenciar a seleção da região mais saliente. Pesquisas nesta área são relativamente novas e os poucos modelos existentes encontram-se em suas fases iniciais. Aqui, nós propomos um novo modelo para atenção visual com modulações bottom-up e top-down. Comparações qualitativas e quantitativas do modelo proposto são realizadas em relação aos mapas de fixação humana e demais modelos estado da arte propostos / Perceiving a complex scene is a quite demanding task for a computer albeit our brain does it efficiently. Evolution has developed ways to optimize our visual system in such a manner that only important parts of the scene undergo scrutiny at a given time. This selection mechanism is named visual attention. Visual attention operates in two modes: bottom-up and top-down. Bottom-up attention is driven by scene-based conspicuities, such as the contrast of colors, orientation, etc. On the other hand, top-down attention is controlled by task, memory, etc. Top-down attention can even modulate the bottom-up mechanism biasing features according to the task. In additional to modulation mechanism taken into account, what is selected from the scene also represents an important part of the selection process. In this scenario, several theories have been proposed and can be gathered in two main lines: space-based attention and object-based attention. Object-based models, instead of only delivering the attention to locations or specific features of the scene, claim that the selection it be performed on object level, it means that the objects are the basic unit of perception. In order to develop models following object-based theories, one needs to consider the integration of a perceptual organization module. This module might segment the objects from the background of the scene based on grouping principles, such as similarity, closeness, etc. Those objects will compete for attention. Several object-based models of visual attention have been proposed in recent years. Research in models of visual attention has mainly focused on the bottom-up guidance of early visual features, disregarding any information about objects. On the other hand, recently works have been conducted regarding the use of the knowledge of the target to influence the computation of the most salient region. The research in this area is rather new and the few existing models are in their early phases. Here, we propose a new visual attention model with both bottom-up and top-down modulations. We provide both qualitative and quantitative comparisons of the proposed model against an ground truth fixation maps and state-of-the-art proposed methods

A superfície da terra foi modificada nos últimos 50 anos mais do que em qualquer outro período da História, mais intensa e rápida nos trópicos pela expansão das frentes de ocupação humana sobre floresta madura. A Amazônia brasileira, caracterizada pela alternância de ciclos econômicos extrativistas, exemplifica esse processo. Entre o abandono de áreas degradadas e a abertura de novas frentes de ocupação, ocorre a regeneração florestal. A floresta secundária tem uma reconhecida importância para o restabelecimento das funções dos ecossistemas e dos estoques de nutrientes perdidos da floresta madura, mas ignorados por muitos anos de taxas oficiais de desmatamento na Amazônia brasileira. Este estudo apresenta uma abordagem utilizando Análise de Imagens Baseada em Objetos Geográficos (GEOBIA) para classificar os estágios de sucessão secundária numa área com cerca de 11.124 km² na região de Santarém (Pará, Brasil). Dentre os resultados, foram produzidas 19 diferentes classificações cobrindo o período 1984 a 2016, que permitiu identificar a redução da floresta madura e da floresta secundária devido à expansão da fronteira agrícola. Outro resultado relevante foi a modelagem de uma árvore de decisão aplicável às imagens de refletância de superfície coletadas pelos satélites LANDSAT, processando esses atributos de classificação em um aplicativo de mineração de dados / The earth surface was modified in the last 50 years more than in any other period of the History, more intense and fast in the tropics by the expansion of human occupation frontiers on the mature forest. The Brazilian Amazon, characterized by alternating extractive economic cycles, exemplifies this process. Between the degraded areas abandonment and the new occupation fronts, forest regeneration takes place. The secondary forest has a recognized importance for the restoration of ecosystem functions and the nutrient stocks lost from the mature forest but ignored for many years of official deforestation rates in the Brazilian Amazon. In this study, an approach using Geographic Object-Based Imaging Analysis (GEOBIA) is presented to classify the stages of secondary succession in an area with near 11,124 km² on Santarém region (Pará State, Brazil). Among the results, 19 different classifications were produced covering the period 1984 to 2016, which allowed identify the reduction of mature forest and secondary forest due to agricultural frontier expansion. Another relevant result was the modeling of a decision tree applicable to surface reflectance images collected by the LANDSAT satellites, processing these classifications attributes in a data mining software

[pt] Neste trabalho, apresentamos dois métodos geostatísticos para modelar estruturas geológicas que exibem características direcionais em uma estrutura de árvore, como leques deltaicos e canais turbidíticos. O primeiro método é um algoritmo geoestatístico multi-ponto chamado simulação baseada em campo de direções (DIR-SIM). A característica direcional da imagem de treinamento é usada para criar um novo objeto que chamamos de campo direcional de treinamento (TDF), que contém a direção em cada ponto da imagem. Este TDF representa a imagem de treinamento em um sentido mais amplo por que tanto a imagem de treinamento quanto a direção seguida pelo reservatório estão contidas nele. Propomos aplicar esse objeto como uma ferramenta fundamental na simulação. O segundo método é uma simulação baseada em objetos chamada SKE-SIM, que usa uma imagem de treinamento para extrair a distribução de parâmetros selecionados para construir o sistema de canais turbidíticos. A idéia baseia-se na premissa de que a imagem de treinamento pode ser bem representada por um objeto unidimensional que chamamos esqueleto. / [en] In this work, we present two geostatistical methods to model geological structures that exhibit directional features in a tree structure, like fan deltas and turbidite channels. The first method is a multiple point geostatistical algorithm called directional field-based simulation (DIR-SIM). The directional feature of the training image is used to create a new object that we call training directional field (TDF), which contains the direction in each point of the image. This TDF represents the training image in a broader sense because both the training image and the direction followed by the reservoir are contained there. We propose to apply this object as a fundamental tool in the simulation. The second method is an object- based simulation called SKE-SIM which uses a training image to extract the distribution of selected parameters to build the turbidite channel system. The idea is based on the premise that the training image can be well represented by a one-dimensional object that we call, skeleton.

A análise de cenas complexas por computadores não é uma tarefa trivial, entretanto, o cérebro humano pode realizar esta função de maneira eficiente. A evolução natural tem desenvolvido formas para otimizar nosso sistema visual de modo que apenas partes importantes da cena sejam analisadas a cada instante. Este mecanismo de seleção é denominado por atenção visual. A atenção visual opera sob dois aspectos: bottom-up e top-down. A atenção bottom-up é dirigida por conspicuidades baseadas na cena, como o contraste de cores, orientação, etc. Por outro lado, a atenção top-down é controlada por tarefas, memórias, etc. A atenção top-down pode ainda modular o mecanismo bottom-up através do enviesamento de determinadas características de acordo com a tarefa. Além do mecanismo de modulação considerado, o que é selecionado a partir da cena também representa uma importante parte para o processo de seleção. Neste cenário, diversas teorias têm sido propostas e podem ser agrupadas em duas linhas principais: atenção baseada no espaço e atenção baseada em objetos. Modelos baseados em objeto, ao invés de apenas direcionar a atenção para locais ou características específicas da cena, requerem que a seleção seja realizada a nível de objeto, significando que os objetos são a unidade básica da percepção. De modo a desenvolver modelos de acordo com a teoria baseada em objetos, deve-se considerar a integração de um módulo de organização perceptual. Este módulo pode segmentar os objetos do fundo da cena baseado em princípios de agrupamento tais como similaridade, proximidade, etc. Esses objetos competirão pela atenção. Diversos modelos de atenção visual baseados em objetos tem sido propostos nos últimos anos. Pesquisas em modelos de atenção visual têm sido desenvolvidas principalmente relacionadas à atenção bottom-up guiadas por características visuais primitivas, desconsiderando qualquer informação sobre os objetos. Por outro lado, trabalhos recentes têm sido realizados em relação ao uso do conhecimento sobre o alvo para influenciar a seleção da região mais saliente. Pesquisas nesta área são relativamente novas e os poucos modelos existentes encontram-se em suas fases iniciais. Aqui, nós propomos um novo modelo para atenção visual com modulações bottom-up e top-down. Comparações qualitativas e quantitativas do modelo proposto são realizadas em relação aos mapas de fixação humana e demais modelos estado da arte propostos / Perceiving a complex scene is a quite demanding task for a computer albeit our brain does it efficiently. Evolution has developed ways to optimize our visual system in such a manner that only important parts of the scene undergo scrutiny at a given time. This selection mechanism is named visual attention. Visual attention operates in two modes: bottom-up and top-down. Bottom-up attention is driven by scene-based conspicuities, such as the contrast of colors, orientation, etc. On the other hand, top-down attention is controlled by task, memory, etc. Top-down attention can even modulate the bottom-up mechanism biasing features according to the task. In additional to modulation mechanism taken into account, what is selected from the scene also represents an important part of the selection process. In this scenario, several theories have been proposed and can be gathered in two main lines: space-based attention and object-based attention. Object-based models, instead of only delivering the attention to locations or specific features of the scene, claim that the selection it be performed on object level, it means that the objects are the basic unit of perception. In order to develop models following object-based theories, one needs to consider the integration of a perceptual organization module. This module might segment the objects from the background of the scene based on grouping principles, such as similarity, closeness, etc. Those objects will compete for attention. Several object-based models of visual attention have been proposed in recent years. Research in models of visual attention has mainly focused on the bottom-up guidance of early visual features, disregarding any information about objects. On the other hand, recently works have been conducted regarding the use of the knowledge of the target to influence the computation of the most salient region. The research in this area is rather new and the few existing models are in their early phases. Here, we propose a new visual attention model with both bottom-up and top-down modulations. We provide both qualitative and quantitative comparisons of the proposed model against an ground truth fixation maps and state-of-the-art proposed methods

Urban land cover mapping represents one of the most important remote sensing applications in the context of rapid global urbanization. In recent years, high resolution spaceborne Polarimetric Synthetic Aperture Radar (PolSAR) has been increasingly used for urban land cover/land-use mapping, since more information could be obtained in multiple polarizations and the collection of such data is less influenced by solar illumination and weather conditions.  The overall objective of this research is to develop effective methods to extract accurate and detailed urban land cover information from spaceborne PolSAR data. Six RADARSAT-2 fine-beam polarimetric SAR and three RADARSAT-2 ultra-fine beam SAR images were used. These data were acquired from June to September 2008 over the north urban-rural fringe of the Greater Toronto Area, Canada. The major landuse/land-cover classes in this area include high-density residential areas, low-density residential areas, industrial and commercial areas, construction sites, roads, streets, parks, golf courses, forests, pasture, water and two types of agricultural crops. In this research, various polarimetric SAR parameters were evaluated for urban land cover mapping. They include the parameters from Pauli, Freeman and Cloude-Pottier decompositions, coherency matrix, intensities of each polarization and their logarithms.  Both object-based and pixel-based classification approaches were investigated. Through an object-based Support Vector Machine (SVM) and a rule-based approach, efficiencies of various PolSAR features and the multitemporal data combinations were evaluated. For the pixel-based approach, a contextual Stochastic Expectation-Maximization (SEM) algorithm was proposed. With an adaptive Markov Random Field (MRF) and a modified Multiscale Pappas Adaptive Clustering (MPAC), contextual information was explored to improve the mapping results. To take full advantages of alternative PolSAR distribution models, a rule-based model selection approach was put forward in comparison with a dictionary-based approach.  Moreover, the capability of multitemporal fine-beam PolSAR data was compared with multitemporal ultra-fine beam C-HH SAR data. Texture analysis and a rule-based approach which explores the object features and the spatial relationships were applied for further improvement. Using the proposed approaches, detailed urban land-cover classes and finer urban structures could be mapped with high accuracy in contrast to most of the previous studies which have only focused on the extraction of urban extent or the mapping of very few urban classes. It is also one of the first comparisons of various PolSAR parameters for detailed urban mapping using an object-based approach. Unlike other multitemporal studies, the significance of complementary information from both ascending and descending SAR data and the temporal relationships in the data were the focus in the multitemporal analysis. Further, the proposed novel contextual analyses could effectively improve the pixel-based classification accuracy and present homogenous results with preserved shape details avoiding over-averaging. The proposed contextual SEM algorithm, which is one of the first to combine the adaptive MRF and the modified MPAC, was able to mitigate the degenerative problem in the traditional EM algorithms with fast convergence speed when dealing with many classes. This contextual SEM outperformed the contextual SVM in certain situations with regard to both accuracy and computation time. By using such a contextual algorithm, the common PolSAR data distribution models namely Wishart, G0p, Kp and KummerU were compared for detailed urban mapping in terms of both mapping accuracy and time efficiency. In the comparisons, G0p, Kp and KummerU demonstrated better performances with higher overall accuracies than Wishart. Nevertheless, the advantages of Wishart and the other models could also be effectively integrated by the proposed rule-based adaptive model selection, while limited improvement could be observed by the dictionary-based selection, which has been applied in previous studies. The use of polarimetric SAR data for identifying various urban classes was then compared with the ultra-fine-beam C-HH SAR data. The grey level co-occurrence matrix textures generated from the ultra-fine-beam C-HH SAR data were found to be more efficient than the corresponding PolSAR textures for identifying urban areas from rural areas. An object-based and pixel-based fusion approach that uses ultra-fine-beam C-HH SAR texture data with PolSAR data was developed. In contrast to many other fusion approaches that have explored pixel-based classification results to improve object-based classifications, the proposed rule-based fusion approach using the object features and contextual information was able to extract several low backscatter classes such as roads, streets and parks with reasonable accuracy. / <p>QC 20121112</p>

Useful information extraction from satellite images for the use of other higher level applications such as road network extraction and update, city planning etc. is a very important and active research area. It is seen that pixel-based techniques becomes insufficient for this task with increasing spatial resolution of satellite imaging sensors day by day. Therefore, the use of object-based techniques becomes indispensable and the segmentation method selection is very crucial for object-based techniques. In this thesis, various segmentation algorithms applied in remote sensing literature are presented and a segmentation process that is based on watersheds and multi-scale segmentation is proposed to use as the segmentation step of an object-based classifier. For every step of the proposed segmentation process, qualitative and quantitative comparisons with alternative approaches are done. The ones which provide best performance are incorporated into the proposed algorithm. Also, an unsupervised segmentation accuracy metric to determine all parameters of the algorithm is proposed. By this way, the proposed segmentation algorithm has become a fully automatic approach. Experiments that are done on a database formed with images taken from Google Earth&reg / software provide promising results.

Following the hierarchical nature of forest resource management, the present work focuses on the natural forest cover at various abstraction levels of details, i.e. categorical land use/land cover (LU/LC) level and a continuous empirical estimation of local operational level. As no single sensor presently covers absolutely all the requirements of the entire levels of forest resource assessment, multisource imagery (i.e. RapidEye, TERRA ASTER and LANDSAT TM), in addition to other data and knowledge have been examined. To deal with this structure, an object-based image analysis (OBIA) approach has been assessed in the destabilized Blue Nile region of Sudan as a potential solution to gather the required information for future forest planning and decision making. Moreover, the spatial heterogeneity as well as the rapid changes observed in the region motivates the inspection for more efficient, flexible and accurate methods to update the desired information. An OBIA approach has been proposed as an alternative analysis framework that can mitigate the deficiency associated with the pixel-based approach. In this sense, the study examines the most popular pixel-based maximum likelihood classifier, as an example of the behavior of spectral classifier toward respective data and regional specifics. In contrast, the OBIA approach analyzes remotely sensed data by incorporating expert analyst knowledge and complimentary ancillary data in a way that somehow simulates human intelligence for image interpretation based on the real-world representation of the features. As the segment is the basic processing unit, various combinations of segmentation criteria were tested to separate similar spectral values into groups of relatively homogeneous pixels. At the categorical subtraction level, rules were developed and optimum features were extracted for each particular class. Two methods were allocated (i.e. Rule Based (RB) and Nearest Neighbour (NN) Classifier) to assign segmented objects to their corresponding classes. Moreover, the study attempts to answer the questions whether OBIA is inherently more precise at fine spatial resolution than at coarser resolution, and how both pixel-based and OBIA approaches can be compared regarding relative accuracy in function of spatial resolution. As anticipated, this work emphasizes that the OBIA approach is can be proposed as an advanced solution particulary for high resolution imagery, since the accuracies were improved at the different scales applied compare with those of pixel-based approach. Meanwhile, the results achieved by the two approaches are consistently high at a finer RapidEye spatial resolution, and much significantly enhanced with OBIA. Since the change in LU/LC is rapid and the region is heterogeneous as well as the data vary regarding the date of acquisition and data source, this motivated the implementation of post-classification change detection rather than radiometric transformation methods. Based on thematic LU/LC maps, series of optimized algorithms have been developed to depict the dynamics in LU/LC entities. Therefore, detailed change “from-to” information classes as well as changes statistics were produced. Furthermore, the produced change maps were assessed, which reveals that the accuracy of the change maps is consistently high. Aggregated to the community-level, social survey of household data provides a comprehensive perspective additionally to EO data. The predetermined hot spots of degraded and successfully recovered areas were investigated. Thus, the study utilized a well-designed questionnaire to address the factors affecting land-cover dynamics and the possible solutions based on local community's perception. At the operational structural forest stand level, the rationale for incorporating these analyses are to offer a semi-automatic OBIA metrics estimates from which forest attribute is acquired through automated segmentation algorithms at the level of delineated tree crowns or clusters of crowns. Correlation and regression analyses were applied to identify the relations between a wide range of spectral and textural metrics and the field derived forest attributes. The acquired results from the OBIA framework reveal strong relationships and precise estimates. Furthermore, the best fitted models were cross-validated with an independent set of field samples, which revealed a high degree of precision. An important question is how the spatial resolution and spectral range used affect the quality of the developed model this was also discussed based on the different sensors examined. To conclude, the study reveals that the OBIA has proven capability as an efficient and accurate approach for gaining knowledge about the land features, whether at the operational forest structural attributes or categorical LU/LC level. Moreover, the methodological framework exhibits a potential solution to attain precise facts and figures about the change dynamics and its driving forces. / Da das Waldressourcenmanagement hierarchisch strukturiert ist, beschäftigt sich die vorliegende Arbeit mit der natürlichen Waldbedeckung auf verschiedenen Abstraktionsebenen, das heißt insbesondere mit der Ebene der kategorischen Landnutzung / Landbedeckung (LU/LC) sowie mit der kontinuierlichen empirischen Abschätzung auf lokaler operativer Ebene. Da zurzeit kein Sensor die Anforderungen aller Ebenen der Bewertung von Waldressourcen und von Multisource-Bildmaterialien (d.h. RapidEye, TERRA ASTER und LANDSAT TM) erfüllen kann, wurden zusätzlich andere Formen von Daten und Wissen untersucht und in die Arbeit mit eingebracht. Es wurde eine objekt-basierte Bildanalyse (OBIA) in einer destabilisierten Region des Blauen Nils im Sudan eingesetzt, um nach möglichen Lösungen zu suchen, erforderliche Informationen für die zukünftigen Waldplanung und die Entscheidungsfindung zu sammeln. Außerdem wurden die räumliche Heterogenität, sowie die sehr schnellen Änderungen in der Region untersucht. Dies motiviert nach effizienteren, flexibleren und genaueren Methoden zu suchen, um die gewünschten aktuellen Informationen zu erhalten. Das Konzept von OBIA wurde als Substitution-Analyse-Rahmen vorgeschlagen, um die Mängel vom früheren pixel-basierten Konzept abzumildern. In diesem Sinne untersucht die Studie die beliebtesten Maximum-Likelihood-Klassifikatoren des pixel-basierten Konzeptes als Beispiel für das Verhalten der spektralen Klassifikatoren in dem jeweiligen Datenbereich und der Region. Im Gegensatz dazu analysiert OBIA Fernerkundungsdaten durch den Einbau von Wissen des Analytikers sowie kostenlose Zusatzdaten in einer Art und Weise, die menschliche Intelligenz für die Bildinterpretation als eine reale Darstellung der Funktion simuliert. Als ein Segment einer Basisverarbeitungseinheit wurden verschiedene Kombinationen von Segmentierungskriterien getestet um ähnliche spektrale Werte in Gruppen von relativ homogenen Pixeln zu trennen. An der kategorische Subtraktionsebene wurden Regeln entwickelt und optimale Eigenschaften für jede besondere Klasse extrahiert. Zwei Verfahren (Rule Based (RB) und Nearest Neighbour (NN) Classifier) wurden zugeteilt um die segmentierten Objekte der entsprechenden Klasse zuzuweisen. Außerdem versucht die Studie die Fragen zu beantworten, ob OBIA in feiner räumlicher Auflösung grundsätzlich genauer ist als eine gröbere Auflösung, und wie beide, das pixel-basierte und das OBIA Konzept sich in einer relativen Genauigkeit als eine Funktion der räumlichen Auflösung vergleichen lassen. Diese Arbeit zeigt insbesondere, dass das OBIA Konzept eine fortschrittliche Lösung für die Bildanalyse ist, da die Genauigkeiten - an den verschiedenen Skalen angewandt - im Vergleich mit denen der Pixel-basierten Konzept verbessert wurden. Unterdessen waren die berichteten Ergebnisse der feineren räumlichen Auflösung nicht nur für die beiden Ansätze konsequent hoch, sondern durch das OBIA Konzept deutlich verbessert. Die schnellen Veränderungen und die Heterogenität der Region sowie die unterschiedliche Datenherkunft haben dazu geführt, dass die Umsetzung von Post-Klassifizierungs- Änderungserkennung besser geeignet ist als radiometrische Transformationsmethoden. Basierend auf thematische LU/LC Karten wurden Serien von optimierten Algorithmen entwickelt, um die Dynamik in LU/LC Einheiten darzustellen. Deshalb wurden für Detailänderung "von-bis"-Informationsklassen sowie Veränderungsstatistiken erstellt. Ferner wurden die erzeugten Änderungskarten bewertet, was zeigte, dass die Genauigkeit der Änderungskarten konstant hoch ist. Aggregiert auf die Gemeinde-Ebene bieten Sozialerhebungen der Haushaltsdaten eine umfassende zusätzliche Sichtweise auf die Fernerkundungsdaten. Die vorher festgelegten degradierten und erfolgreich wiederhergestellten Hot Spots wurden untersucht. Die Studie verwendet einen gut gestalteten Fragebogen um Faktoren die die Dynamik der Änderung der Landbedeckung und mögliche Lösungen, die auf der Wahrnehmung der Gemeinden basieren, anzusprechen. Auf der Ebene des operativen strukturellen Waldbestandes wird die Begründung für die Einbeziehung dieser Analysen angegeben um semi-automatische OBIA Metriken zu schätzen, die aus dem Wald-Attribut durch automatisierte Segmentierungsalgorithmen in den Baumkronen abgegrenzt oder Cluster von Kronen Ebenen erworben wird. Korrelations- und Regressionsanalysen wurden angewandt, um die Beziehungen zwischen einer Vielzahl von spektralen und strukturellen Metriken und den aus den Untersuchungsgebieten abgeleiteten Waldattributen zu identifizieren. Die Ergebnisse des OBIA Rahmens zeigen starke Beziehungen und präzise Schätzungen. Die besten Modelle waren mit einem unabhängigen Satz von kreuz-validierten Feldproben ausgestattet, welche hohe Genauigkeiten ergaben. Eine wichtige Frage ist, wie die räumliche Auflösung und die verwendete Bandbreite die Qualität der entwickelten Modelle auch auf der Grundlage der verschiedenen untersuchten Sensoren beeinflussen. Schließlich zeigt die Studie, dass OBIA in der Lage ist, als ein effizienter und genauer Ansatz Kenntnisse über die Landfunktionen zu erlangen, sei es bei operativen Attributen der Waldstruktur oder auch auf der kategorischen LU/LC Ebene. Außerdem zeigt der methodischen Rahmen eine mögliche Lösung um präzise Fakten und Zahlen über die Veränderungsdynamik und ihre Antriebskräfte zu ermitteln.

Multitemporal SAR images have been increasingly used for the detection of different types of environmental changes. The detection of urban changes using SAR images is complicated due to the complex mixture of the urban environment and the special characteristics of SAR images, for example, the existence of speckle. This thesis investigates urban change detection using multitemporal SAR images with the following specific objectives: (1) to investigate unsupervised change detection, (2) to investigate effective methods for reduction of the speckle effect in change detection, (3) to investigate spatio-contextual change detection, (4) to investigate object-based unsupervised change detection, and (5) to investigate a new technique for object-based change image generation. Beijing and Shanghai, the largest cities in China, were selected as study areas. Multitemporal SAR images acquired by ERS-2 SAR and ENVISAT ASAR sensors were used for pixel-based change detection. For the object-based approaches, TerraSAR-X images were used. In Paper I, the unsupervised detection of urban change was investigated using the Kittler-Illingworth algorithm. A modified ratio operator that combines positive and negative changes was used to construct the change image. Four density function models were tested and compared. Among them, the log-normal and Nakagami ratio models achieved the best results. Despite the good performance of the algorithm, the obtained results suffer from the loss of fine geometric detail in general. This was a consequence of the use of local adaptive filters for speckle suppression. Paper II addresses this problem using the nonlocal means (NLM) denoising algorithm for speckle suppression and detail preservation. In this algorithm, denoising was achieved through a moving weighted average. The weights are a function of the similarity of small image patches defined around each pixel in the image. To decrease the computational complexity, principle component analysis (PCA) was used to reduce the dimensionality of the neighbourhood feature vectors. Simple methods to estimate the number of significant PCA components to be retained for weights computation and the required noise variance were proposed. The experimental results showed that the NLM algorithm successfully suppressed speckle effects, while preserving fine geometric detail in the scene. The analysis also indicates that filtering the change image instead of the individual SAR images was effective in terms of the quality of the results and the time needed to carry out the computation. The Markov random field (MRF) change detection algorithm showed limited capacity to simultaneously maintain fine geometric detail in urban areas and combat the effect of speckle. To overcome this problem, Paper III utilizes the NLM theory to define a nonlocal constraint on pixels class-labels. The iterated conditional mode (ICM) scheme for the optimization of the MRF criterion function is extended to include a new step that maximizes the nonlocal probability model. Compared with the traditional MRF algorithm, the experimental results showed that the proposed algorithm was superior in preserving fine structural detail, effective in reducing the effect of speckle, less sensitive to the value of the contextual parameter, and less affected by the quality of the initial change map. Paper IV investigates object-based unsupervised change detection using very high resolution TerraSAR-X images over urban areas. Three algorithms, i.e., Kittler-Illingworth, Otsu, and outlier detection, were tested and compared. The multitemporal images were segmented using multidate segmentation strategy. The analysis reveals that the three algorithms achieved similar accuracies. The achieved accuracies were very close to the maximum possible, given the modified ratio image as an input. This maximum, however, was not very high. This was attributed, partially, to the low capacity of the modified ratio image to accentuate the difference between changed and unchanged areas. Consequently, Paper V proposes a new object-based change image generation technique. The strong intensity variations associated with high resolution and speckle effects render object mean intensity unreliable feature. The modified ratio image is, therefore, less efficient in emphasizing the contrast between the classes. An alternative representation of the change data was proposed. To measure the intensity of change at the object in isolation of disturbances caused by strong intensity variations and speckle effects, two techniques based on the Fourier transform and the Wavelet transform of the change signal were developed. Qualitative and quantitative analyses of the result show that improved change detection accuracies can be obtained by classifying the proposed change variables. / <p>QC 20150529</p>

Thesis (MSc)--Stellenbosch University, 2012. / ENGLISH ABSTRACT: Supervised classifiers are commonly used in remote sensing to extract land cover information. They are, however, limited in their ability to cost-effectively produce sufficiently accurate land cover maps. Various factors affect the accuracy of supervised classifiers. Notably, the number of available training samples is known to significantly influence classifier performance and to obtain a sufficient number of samples is not always practical. The support vector machine (SVM) does perform well with a limited number of training samples. But little research has been done to evaluate SVM’s performance for geographical object-based image analysis (GEOBIA). GEOBIA also allows the easy integration of additional features into the classification process, a factor which may significantly influence classification accuracies. As such, two experiments were developed and implemented in this research. The first compared the performances of object-based SVM, maximum likelihood (ML) and nearest neighbour (NN) classifiers using varying training set sizes. The effect of feature dimensionality on classifier accuracy was investigated in the second experiment. A SPOT 5 subscene and a four-class classification scheme were used. For the first experiment, training set sizes ranging from 4-20 per land cover class were tested. The performance of all the classifiers improved significantly as the training set size was increased. The ML classifier performed poorly when few (<10 per class) training samples were used and the NN classifier performed poorly compared to SVM throughout the experiment. SVM was the superior classifier for all training set sizes although ML achieved competitive results for sets of 12 or more training samples per class. Training sets were kept constant (20 and 10 samples per class) for the second experiment while an increasing number of features (1 to 22) were included. SVM consistently produced superior classification results. SVM and NN were not significantly (negatively) affected by an increase in feature dimensionality, but ML’s ability to perform under conditions of large feature dimensionalities and few training areas was limited. Further investigations using a variety of imagery types, classification schemes and additional features; finding optimal combinations of training set size and number of features; and determining the effect of specific features should prove valuable in developing more costeffective ways to process large volumes of satellite imagery. KEYWORDS Supervised classification, land cover, support vector machine, nearest neighbour classification maximum likelihood classification, geographic object-based image analysis / AFRIKAANSE OPSOMMING: Gerigte klassifiseerders word gereeld aangewend in afstandswaarneming om inligting oor landdekking te onttrek. Sulke klassifiseerders het egter beperkte vermoëns om akkurate landdekkingskaarte koste-effektief te produseer. Verskeie faktore het ʼn uitwerking op die akkuraatheid van gerigte klassifiseerders. Dit is veral bekend dat die getal beskikbare opleidingseenhede ʼn beduidende invloed op klassifiseerderakkuraatheid het en dit is nie altyd prakties om voldoende getalle te bekom nie. Die steunvektormasjien (SVM) werk goed met beperkte getalle opleidingseenhede. Min navorsing is egter gedoen om SVM se verrigting vir geografiese objek-gebaseerde beeldanalise (GEOBIA) te evalueer. GEOBIA vergemaklik die integrasie van addisionele kenmerke in die klassifikasie proses, ʼn faktor wat klassifikasie akkuraathede aansienlik kan beïnvloed. Twee eksperimente is gevolglik ontwikkel en geïmplementeer in hierdie navorsing. Die eerste eksperiment het objekgebaseerde SVM, maksimum waarskynlikheids- (ML) en naaste naburige (NN) klassifiseerders se verrigtings met verskillende groottes van opleidingstelle vergelyk. Die effek van kenmerkdimensionaliteit is in die tweede eksperiment ondersoek. ʼn SPOT 5 subbeeld en ʼn vier-klas klassifikasieskema is aangewend. Opleidingstelgroottes van 4-20 per landdekkingsklas is in die eerste eksperiment getoets. Die verrigting van die klassifiseerders het beduidend met ʼn toename in die grootte van die opleidingstelle verbeter. ML het swak presteer wanneer min (<10 per klas) opleidingseenhede gebruik is en NN het, in vergelyking met SVM, deurgaans swak presteer. SVM het die beste presteer vir alle groottes van opleidingstelle alhoewel ML kompeterend was vir stelle van 12 of meer opleidingseenhede per klas. Die grootte van die opleidingstelle is konstant gehou (20 en 10 eenhede per klas) in die tweede eksperiment waarin ʼn toenemende getal kenmerke (1 tot 22) toegevoeg is. SVM het deurgaans beter klassifikasieresultate gelewer. SVM en NN was nie beduidend (negatief) beïnvloed deur ʼn toename in kenmerkdimensionaliteit nie, maar ML se vermoë om te presteer onder toestande van groot kenmerkdimensionaliteite en min opleidingsareas was beperk. Verdere ondersoeke met ʼn verskeidenheid beelde, klassifikasie skemas en addisionele kenmerke; die vind van optimale kombinasies van opleidingstelgrootte en getal kenmerke; en die bepaling van die effek van spesifieke kenmerke sal waardevol wees in die ontwikkelling van meer koste effektiewe metodes om groot volumes satellietbeelde te prosesseer. TREFWOORDE Gerigte klassifikasie, landdekking, steunvektormasjien, naaste naburige klassifikasie, maksimum waarskynlikheidsklassifikasie, geografiese objekgebaseerde beeldanalise

Submitted by Emerson Leão Júnior null (emerson.leaojr@gmail.com) on 2017-12-05T18:07:16Z No. of bitstreams: 1 leao_ej_me_prud.pdf: 4186679 bytes, checksum: ee186b23411343c3e2d782d622226699 (MD5) / Approved for entry into archive by ALESSANDRA KUBA OSHIRO null (alessandra@fct.unesp.br) on 2017-12-06T10:52:22Z (GMT) No. of bitstreams: 1 leaojunior_e_me_prud.pdf: 4186679 bytes, checksum: ee186b23411343c3e2d782d622226699 (MD5) / Made available in DSpace on 2017-12-06T10:52:22Z (GMT). No. of bitstreams: 1 leaojunior_e_me_prud.pdf: 4186679 bytes, checksum: ee186b23411343c3e2d782d622226699 (MD5) Previous issue date: 2017-04-25 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / Considerando o cenário durante a crise hídrica de 2014 e a situação crítica dos reservatórios do sistema Cantareira no estado de São Paulo, este estudo realizado no reservatório Jaguari-Jacareí, consistiu na extração de informações a partir de imagens multiespectrais e análise da qualidade da informação relacionada com a acurácia no cálculo do volume de água do reservatório. Inicialmente, a superfície do espelho d’água foi obtida pela classificação da cobertura da terra a partir de imagens multiespectrais RapidEye tomadas antes e durante a crise hídrica (2013 e 2014, respectivamente), utilizando duas abordagens distintas: classificação orientada a objeto (Object-based Image Analysis - OBIA) e classificação baseada em pixel (Support Vector Machine – SVM). A acurácia do usuário por classe permitiu expressar o erro para detectar a superfície do espelho d’água para cada abordagem de classificação de 2013 e 2014. O segundo componente da estimação do volume foi a representação do relevo submerso, que considerou duas fontes de dados na construção do modelo numérico do terreno (MNT): dados topográficos provenientes de levantamento batimétrico disponibilizado pela Sabesp e o modelo de superfície AW3D30 (ALOS World 3D 30m mesh), para complementar a informação não disponível além da cota 830,13 metros. A comparação entre as duas abordagens de classificação dos tipos de cobertura da terra do entorno do reservatório Jaguari-Jacareí mostrou que SVM resultou em indicadores de acurácia ligeiramente superiores à OBIA, para os anos de 2013 e 2014. Em relação à estimação de volume do reservatório, incorporando a informação do nível de água divulgado pela Sabesp, a abordagem SVM apresentou menor discrepância relativa do que OBIA. Apesar disso, a qualidade da informação produzida na estimação de volume, resultante da propagação da variância associada aos dados envolvidos no processo, ambas as abordagens produziram valores similares de incerteza, mas com uma sutil superioridade de OBIA, para alguns dos cenários avaliados. No geral, os métodos de classificação utilizados nesta dissertação produziram informação acurada e adequada para o monitoramento de recursos hídricos e indicou que a abordagem SVM teve um desempenho sutilmente superior na classificação dos tipos de cobertura da terra, na estimação do volume e em alguns dos cenários considerados na propagação da incerteza. / This study aims to extract information from multispectral images and to analyse the information quality in the water volume estimation of Jaguari-Jacareí reservoir. The presented study of changes in the volume of the Jaguari-Jacareí reservoir was motivated by the critical situation of the reservoirs from Cantareira System in São Paulo State caused by water crisis in 2014. Reservoir area was extracted from RapidEye multispectral images acquired before and during the water crisis (2013 and 2014, respectively) through land cover classification. Firstly, the image classification was carried out in two distinct approaches: object-based (Object-based Image Analysis - OBIA) and pixel-based (Support Vector Machine - SVM) method. The classifications quality was evaluated through thematic accuracy, in which for every technique the user accuracy allowed to express the error for the class representing the water in 2013 and 2014. Secondly, we estimated the volume of the reservoir’s water body, using the numerical terrain model generated from two additional data sources: topographic data from a bathymetric survey, available from Sabesp, and the elevation model AW3D30 (to complement the information in the area where data from Sabesp was not available). When compare the two classification techniques, it was found that in the image classification, SVM performance slightly overcame the OBIA classification technique for 2013 and 2014. In the volume calculation considering the water level estimated from the generated DTM, the result obtained by SVM approach was better in 2013, whereas OBIA approach was more accurate in 2014. Considering the quality of the information produced in the volume estimation, both approaches presented similar values of uncertainty, with the OBIA method slightly less uncertain than SVM. In conclusion, the classification methods used in this dissertation produced accurate information to monitor water resource, but SVM had a subtly superior performance in the classification of land cover types, volume estimation and some of the scenarios considered in the propagation of uncertainty.