Showing papers by "Peter Reinartz published in 2012"

Semiglobal Matching Results on the ISPRS Stereo Matching Benchmark

Abstract: . Digital surface models can be efficiently generated with automatic image matching from optical stereo images. The Working Group 4 of Commission I on "Geometric and Radiometric Modelling of Optical Spaceborne Sensors" provides a matching benchmark dataset with several stereo data sets from high and very high resolution space borne stereo sensors at http://www.commission1.isprs.org/wg4/ . The selected regions are in Catalonia, Spain, and include three test areas, covering city areas, rural areas and forests in flat and medium undulated terrain as well as steep mountainous terrain. In this paper, digital surface models (DSM) are derived from the Cartosat-1 and Worldview-1 datasets using Semiglobal Matching. The resulting DSM are evaluated against the first pulse returns of the LIDAR reference dataset provided by the Institut Cartogr`afic de Catalunya (ICC), using robust accuracy measures.

Performance Evaluation for 3-D City Model Generation of Six Different DSMs From Air- and Spaceborne Sensors

Abstract: Since remote sensing provides more and more sensors and techniques to accumulate data on urban regions, three-dimensional representations of these complex environments gained much interest for various applications. In order to obtain three-dimensional representations, one of the most practical ways is to generate Digital Surface Models (DSMs) using very high resolution remotely sensed images from two or more viewing directions, or by using LIDAR sensors. Due to occlusions, matching errors and interpolation techniques these DSMs do not exhibit completely steep walls, and in order to obtain real three-dimensional urban models including objects like buildings from these DSMs, advanced methods are needed. A novel approach based on building shape detection, height estimation, and rooftop reconstruction is proposed to achieve realistic three-dimensional building representations. Our automatic approach consists of three main modules as; detection of complex building shapes, understanding rooftop type, and three-dimensional building model reconstruction based on detected shape and rooftop type. Besides the development of the methodology, the goal is to investigate the applicability and accuracy which can be accomplished in this context for different stereo sensor data. We use DSMs of Munich city which are obtained from different satellite (Cartosat-1, Ikonos, WorldView-2) and airborne sensors (3K camera, HRSC, and LIDAR). The paper later focuses on a quantitative comparisons of the outputs from the different multi-view sensors for a better understanding of qualities, capabilities and possibilities for applications. Results look very promising even for the DSMs derived from satellite data.

Automated Georeferencing of Optical Satellite Data with Integrated Sensor Model Improvement

Abstract: The geometric processing of remotely sensed image data is one of the key issues in data interpretation, added value product generation, and multi-source data integration. Although optical satellite data can be orthorectified without the use of Ground Control Points (GCP) to absolute geometric accuracies of some meters up to several hundred meters depending on the satellite mission, there is still a need to improve the geometric accuracy by using GCP. The manual measurement of GCP is time consuming work, and leads, especially for larger data sets with hundreds of satellite images, to a cost and time ineffective workload. To overcome these shortcomings, an autonomous processing chain to georeference and orthorectify optical satellite data is proposed which uses reference data and digital elevation models to generate GCP and to improve sensor model parameters (namely for rigorous and universal sensor models) for a series of optical Earth observation satellite systems. Using a restrictive blunder removal strategy, the proposed procedure leads to high quality orthorectified products or at least to a geometrically consistent data set in terms of relative accuracy. The geometric processing chain is validated using SPOT-4 HRVIR, SPOT-5 HRG, IRS-P6 LISS III, and ALOS AVNIR-2 optical sensor data, for which a huge amount of satellite data (3,200 scenes) has been processed. Relative and absolute geometric accuracies of approximately half the pixel size (linear Root Mean Square Error) are achieved.

Hyperspectral image resolution enhancement based on spectral unmixing and information fusion

Abstract: . Hyperspectral imaging sensors exibit high spectral resolution, but normally low spatial resolution. This leads to spectral signatures of pixels originating from different object types. Such pixels are called mixed pixels. Spectral unmixing methods can be employed to estimate the fractions of reflected light from the different objects within the pixel area. However, spectral unmixing does not provide any spatial information about the sources and therefore additional information is needed to precisely locate the sources. In order to restore the spatial information of hyperspectral images we propose a hyperspectral and multispectral image fusion method based on spectral unmixing. The algorithm is tested with HyMAP image data consisting of 125 spectral bands and a simulated multispectral image consisting of 8 bands.

Low-cost optical Camera System for real-time Mapping Applications

Abstract: Die Beobachtung von Naturkatastrophen, Grosereignissen und Unfallen mit flugzeuggestutzten optischen Sensoren in Echtzeit ist ein derzeit wichtiges Thema in Forschung und Entwicklung. In diesem Zusammenhang wird die Leistungsfahigkeit von preisgunstigen Kamerasystemen fur Echtzeitanwendungen in Hinblick auf geometrische Genauigkeit, radiometrische Eigenschaften und Prozessierungszeiten evaluiert. Der Schwerpunkt liegt bei der Analyse der geometrischen Stabilitat von preisgunstigen Kameras im langjahrigen Betrieb und den Grenzen der direkten Georeferenzierung. Weiterhin wird eine echtzeitfahige Prozessierungskette mit einer GPU (Graphical Processing Unit) basierten Orthorektifizierungsmethode fur eine maximal mogliche Aufnahmerate von 5Hz vorgestellt.

Dsm based orientation of large stereo satellite image blocks

Abstract: High resolution stereo satellite imagery is well suited for the creation of digital surface models (DSM) A system for highly automated and operational DSM and orthoimage generation based on CARTOSAT-1 imagery is presented, with emphasis on fully automated georeferencing The proposed system processes level-1 stereo scenes using the rational polynomial coefficients (RPC) universal sensor model The RPC are derived from orbit and attitude information and have a much lower accuracy than the ground resolution of approximately 25 m In order to use the images for orthorectification or DSM generation, an affine RPC correction is required In this paper, GCP are automatically derived from lower resolution reference datasets (Landsat ETMp Geocover and SRTM DSM) The traditional method of collecting the lateral position from a reference image and interpolating the corresponding height from the DEM ignores the higher lateral accuracy of the SRTM dataset Our method avoids this drawback by using a RPC correction based on DSM alignment, resulting in improved geolocation of both DSM and ortho images Scene based method and a bundle block adjustment based correction are developed and evaluated for a test site covering the nothern part of Italy, for which 405 Cartosat-1 Stereopairs are available Both methods are tested against independent ground truth Checks against this ground truth indicate a lateral error of 10 meters

Complex building description and extraction based on Hough transformation and cycle detection

Abstract: A simple but robust approach for complex building description and extraction from high-resolution remotely sensed imagery based on graph-based shape representation is proposed. Classical approaches for building extraction usually involve a complex grouping process of low-level primitive features and are not robust in the presence of noise. To overcome these drawbacks, this approach presents an efficient and robust solution by integrating edges and regions. First, a region segmentation method is applied to obtain the approximate shape of the building. Second, Hough transformation is employed to derive the two perpendicular line sets corresponding to the building boundary. Third, a subset of the intersectional nodes of the two line sets is utilized to construct a building structural graph, based on the analysis of grey value difference between the two sides of each line segment. Finally, a graph search algorithm is performed to retrieve all the cycles in the structural graph. The cycle corresponding to the building boundary is identified as the final building outline on the basis of its area. Two experiments were carried out to evaluate and validate this approach and experimental results confirm its effectiveness and robustness.

Combined Edge Segment Texture Analysis for the Detection of Damaged Buildings in Crisis Areas

Abstract: This paper describes the results of a new combined method that consists of a cooperative approach of several different algorithms for automated change detection. These methods are based on isotropic frequency filtering, spectral and texture analysis, and segmentation. For the frequency analysis, different band pass filters are applied to identify the relevant frequency information for change detection. After transforming the multitemporal images using a fast Fourier transform and applying the most suitable band pass filter to extract changed structures, we apply an edge detection algorithm in the spatial domain. For the texture analysis, we calculate the parameters energy and homogeneity for the multitemporal datasets. Then a principal component analysis is applied to the new multispectral texture images and subtracted to get the texture change information. This method can be combined with spectral information and prior segmentation of the image data as well as with morphological operations for a final binary change result. A rule-based combination of the change algorithms is applied to calculate the probability of change for a particular location. This Combined Edge Segment Texture (CEST) method was tested with high-resolution remote-sensing images of the crisis area in Darfur (Sudan). Our results were compared with several standard algorithms for automated change detection, such as image difference, image ratio, principal component analysis, multivariate alteration detection (MAD) and post classification change detection. CEST showed superior accuracy compared to standard methods.

Analysis and selection of pan-sharpening assessment measures

Abstract: Pan-sharpening of remote sensing multispectral imagery directly influences the accuracy of interpretation, classification, and other data mining methods. Different tasks of multispectral image analysis and processing require specific properties of input pan-sharpened multispectral data such as spectral and spatial consistency, complexity of the pan-sharpening method, and other properties. The quality of a pan-sharpened image is assessed using quantitative measures. Generally, the quantitative measures for pan-sharpening assessment are taken from other topics of image processing (e.g., image similarity indexes), but the applicability basis of these measures (i.e., whether a measure provides correct and undistorted assessment of pan-sharpened imagery) is not checked and proven. For example, should (or should not) a quantitative measure be used for pan-sharpening assessment is still an open research topic. Also, there is a chance that some measures can provide distorted results of the quality assessment and the suitability of these quantitative measures as well as the application for pan-sharpened imagery assessment is under question. The aim of the authors is to perform statistical analysis of widely employed measures for remote sensing imagery pan-sharpening assessment and to show which of the measures are the most suitable for use. To find and prove which measures are the most suitable, sets of multispectral images are processed by the general fusion framework method (GFF) with varying parameters. The GFF is a type of general image fusion method. Variation of the method parameter set values allows one to produce imagery data with predefined quality (i.e., spatial and spectral consistency) for further statistical analysis of the assessment measures. The use of several main multispectral sensors (Landsat 7ETM+, IKONOS, and WorldView-2) imagery allows one to assess and compare available quality assessment measures and illustrate which of them are most suitable for each satellite.

Matching of high-resolution optical data to a shaded DEM

Abstract: One of the first essential steps in the analysis of optical satellite imagery is orthorectification of the images. Orthorectification without ground control points (GCPs) using only the ephemeris and attitude data (or rational polynomial functions) provided by the satellite operator leads to an absolute accuracy of about 10 m to 1 km (depending on the satellite and available metadata). The orientation can be improved by measuring GCPs with well-known coordinates in the images. In this article, an alternative method to obtain GCPs from an existing digital elevation model (DEM) is described and assessed. Since at least the Shuttle Radar Topography Mission DEM is available nearly worldwide with absolute horizontal accuracy of some metres, these DEMs may serve as a valuable additional source for the generation of GCPs. Furthermore, several planned and ongoing missions will increase the availability and accuracy of DEMs or stereo imagery, respectively, e.g. ALOS-PRISM, TanDEM-X, etc. Results for several test a...

Fusing stereo and multispectral data from WorldView-2 for urban modeling

Abstract: Using the capability of WorldView-2 to acquire very high resolution (VHR) stereo imagery together with as much as eight spectral channels allows the worldwide monitoring of any built up areas, like cities in evolving states. In this paper we show the benefit of generating a high resolution digital surface model (DSM) from multi-view stereo data (PAN) and fusing it with pan sharpened multi-spectral data to arrive at very detailed information in city areas. The fused data allow accurate object detection and extraction and by this also automated object oriented classification and future change detection applications. The methods proposed in this paper exploit the full range of capacities provided by WorldView-2, which are the high agility to acquire a minimum of two but also more in-orbit-images with small stereo angles, the very high ground sampling distance (GSD) of about 0.5 m and also the full usage of the standard four multispectral channels blue, green, red and near infrared together with the additional provided channels special to WorldView-2: coastal blue, yellow, red-edge and a second near infrared channel. From the very high resolution stereo panchromatic imagery a so called height map is derived using the semi global matching (SGM) method developed at DLR. This height map fits exactly on one of the original pan sharpened images. This in turn is used for an advanced rule based fuzzy spectral classification. Using these classification results the height map is corrected and finally a terrain model and an improved normalized digital elevation model (nDEM) generated. Fusing the nDEM with the classified multispectral imagery allows the extraction of urban objects like like buildings or trees. If such datasets from different times are generated the possibility of an expert object based change detection (in quasi 3D space) and automatic surveillance will become possible.

On the use of overcomplete dictionaries for spectral unmixing

Abstract: Hyperspectral unmixing is a sub pixel classification method which aims at recovering fraction and type of materials mixed in a single pixel. This work addresses the unmixing problem from the compressive sensing point of view by using overcomplete dictionaries enabling automatization of the process. However, overcomplete dictionaries of spectra are highly coherent which might confuse the final unmixing result. To deal with this problem we propose the use of differentiated spectra for coherence reduction. In this paper we study the approximation error for the proposed method as well as the correctness of the material detection.

Kalman filter based feature analysis for tracking people from airborne images

Abstract: Recently, analysis of man events in real-time using computer vision techniques became a very important research field. Especially, understanding motion of people can be helpful to prevent unpleasant conditions. Understanding behavioral dynamics of people can also help to estimate future states of underground passages, shopping center like public entrances, or streets. In order to bring an automated solution to this problem, we propose a novel approach using airborne image sequences. Although airborne image resolutions are not enough to see each person in detail, we can still notice a change of color components in the place where a person exists. Therefore, we propose a color feature detection based probabilistic framework in order to detect people automatically. Extracted local features behave as observations of the probability density function (pdf) of the people locations to be estimated. Using an adaptive kernel density estimation method, we estimate the corresponding pdf. First, we use estimated pdf to detect boundaries of dense crowds. After that, using background information of dense crowds and previously extracted local features, we detect other people in non-crowd regions automatically for each image in the sequence. We benefit from Kalman filtering to track motion of detected people. To test our algorithm, we use a stadium entrance image data set taken from airborne camera system. Our experimental results indicate possible usage of the algorithm in real-life man events. We believe that the proposed approach can also provide crucial information to police departments and crisis management teams to achieve more detailed observations of people in large open area events to prevent possible accidents or unpleasant conditions.

Detecting changes between a DSM and a high resolution SAR image with the support of simulation based separation of urban scenes

Abstract: As a SAR image is often the only available data in crisis situations, e.g. after an earthquake, a change analysis of the SAR image with previously acquired data may enable a fast analysis of the damage caused by the disaster. This paper presents a method for change analysis of the simulated appearance of a digital surface model (DSM) and a SAR image. A simulated SAR image is generated using the DSM and is separated into four masks indicating double bounce reflection, layover, shadow areas and ground reflection. Temporal changes of the corresponding areas in the real SAR image are analysed using pixel-based methods. Finally, the change detection results for the four masks are combined in order to obtain a robust change analysis result for single buildings. In this regard, the application of the proposed concept is presented for the city centre of Munich using TerraSAR-X data.

Low-cost optical camera system for disaster monitoring

Abstract: Real-time monitoring of natural disasters, mass events, and large accidents with airborne optical sensors is an ongoing topic in research and development. Airborne monitoring is used as a complemental data source with the advantage of flexible data acquisition and higher spatial resolution compared to optical satellite data. In cases of disasters or mass events, optical high resolution image data received directly after acquisition are highly welcomed by security related organizations like police and rescue forces. Low-cost optical camera systems are suitable for real-time applications as the accuracy requirements can be lowered in return for faster processing times. In this paper, the performance of low-cost camera systems for real-time mapping applications is exemplarily evaluated based on already existing sensor systems operated at German Aerospace Center (DLR). Focus lies next to the geometrical and radiometric performance on the real time processing chain which includes image processors, thematic processors for automatic traffic extraction and automatic person tracking, data downlink to the ground station, and further processing and distribution on the ground. Finally, a concept for a national airborne rapid mapping service based on the low-cost hardware is proposed.

Evaluation of selected features for car detection in aerial images

Abstract: The extraction of vehicles from aerial images provides a wide area traffic situation within a short time. Applications for the gathered data are various and reach from smart routing in the case of congestions to usability validation of roads in the case of disasters. The challenge of the vehicle detection task is finding adequate features which are capable to separate cars from other objects; especially those that look similar. We present an experiment where selected features show their ability of car detection. Precisely, Haar-like and HoG features are utilized and passed to the AdaBoost algorithm for calculating the final detector. Afterwards the classifying power of the features is accurately analyzed and evaluated. The tests a carried out on aerial data from the inner city of Munich, Germany and include small inner city roads with rooftops close by which raise the complexity factor.

Synergetic use of TerraSAR-X and Radarsat-2 time series data for identification and characterization of grassland types - a case study in Southern Bavaria, Germany

Abstract: In the context of global change, alteration of landscapes and loss of biodiversity, the monitoring of habitats, vegetation types and their changes have become extraordinary important. In this paper, first results from a study that analyses the differentiability of NATURA2000 habitats and HNV grassland with imaging radar data are presented. Therefore, Kennaugh elements derived from TerraSAR-X and Radarsat-2 dual pol (VV/VH) time series data are used, both separately and in combination, to model the distribution of these classes with the Maximum-Entropy principle. The preliminary results show that the multi-frequency approach enables - compared to single frequency analyses - a finer differentiation between scatterers in the size of 3–6 cm (e.g. 7120, 7230 and HNV grassland).

Rapid change detection algorithm for disaster management

Abstract: This paper focuses on change detection applications in areas where catastrophic events took place which resulted in rapid destruction especially of manmade objects. Standard methods for automated change detection prove not to be sufficient; therefore a new method was developed and tested. The presented method allows a fast detection and visualization of change in areas of crisis or catastrophes. While often new methods of remote sensing are developed without user oriented aspects, organizations and authorities are not able to use these methods because of absence of remote sensing know how. Therefore a semi-automated procedure was developed. Within a transferable framework, the developed algorithm can be implemented for a set of remote sensing data among different investigation areas. Several case studies are the base for the retrieved results. Within a coarse dividing into statistical parts and the segmentation in meaningful objects, the framework is able to deal with different types of change. By means of an elaborated Temporal Change Index (TCI) only panchromatic datasets are used to extract areas which are destroyed, areas which were not affected and in addition areas where rebuilding has already started.

Automatic interpretation of high resolution sar images: first results of sar image simulation for single buildings

Abstract: Due to the all-weather data acquisition capabilities, high resolution space borne Synthetic Aperture Radar (SAR) plays an important role in remote sensing applications like change detection However, because of the complex geometric mapping of buildings in urban areas, SAR images are often hard to interpret SAR simulation techniques ease the visual interpretation of SAR images, while fully automatic interpretation is still a challenge This paper presents a method for supporting the interpretation of high resolution SAR images with simulated radar images using a LiDAR digital surface model (DSM) Line features are extracted from the simulated and real SAR images and used for matching A single building model is generated from the DSM and used for building recognition in the SAR image An application for the concept is presented for the city centre of Munich where the comparison of the simulation to the TerraSAR-X data shows a good similarity Based on the result of simulation and matching, special features (eg like double bounce lines, shadow areas etc) can be automatically indicated in SAR image

Region based forest change detection from cartosat-1 stereo imagery

Abstract: Tree height is a fundamental parameter for describing the forest situation and changes The latest development of automatic Digital Surface Model (DSM) generation techniques allows new approaches of forest change detection from satellite stereo imagery This paper shows how DSMs can support the change detection in forest area A novel region based forest change detection method is proposed using single-channel CARTOSAT-1 stereo imagery In the first step, DSMs from two dates are generated based on automatic matching technology After co-registration and normalising by using LiDAR data, the mean-shift segmentation is applied to the original pan images, and the images of both dates are classified to forest and non-forest areas by analysing their histograms and height differences In the second step, a rough forest change detection map is generated based on the comparison of the two forest map Then the GLCM texture from the nDSM and the Cartosat-1 images of the resulting regions are analyzed and compared, the real changes are extracted by SVM based classification

Automatic population counts for improved wildlife management using aerial photography

Abstract: For effective conservation management, it is very important to provide accurate estimates of animal populations with certain time intervals. So far many studies are performed visually/manually which requires much time and is prone to errors. Besides, only a limited area can be considered for counting because of the effort required. In order to bring a new solution to this problem, herein we propose a novel approach for counting animals from aerial images by using computer vision techniques. To do so, we apply a probabilistic framework on local features in the image whose spectral reflectance differs from the surrounding region. We use mean shift segmentation and obtain probability density function (pdf) to detect focus of attention regions (FOA). Finally, we benefit from graph theory to detect segments which should represent animals. We test the feasibility of the proposed approach using aerial images of varying quality and angles (including orthogonal time lapse photography) from several different terrestrial ecosystems. Monitored species include birds and mammals. The algorithms successfully detect and count animals and provide a replicable and objective method for estimating animal abundance, however the methodology still requires estimates of error to be incorporated. This approach highlights how technical innovations in remote sensing can provide valuable information for conservation management.

Sparse Approximation, Coherence and Use of Derivatives in Hyperspectral Unmixing

Abstract: Recently, it has been shown that the spectral unmixing can be regarded as a sparse approximation problem. In our studies we employ predefined dictionaries containing the measured spectra of different materials in a hyperspectral image, where for each pixel the abundance vector can be estimated solving the $\ell_1$ optimization problem. This results in an automation of the unmixing procedure and enables using complex overcomplete dictionaries. However, the reflectance spectra of most materials are highly coherent and this could result in confusion in the mixture estimation. In this work we present a novel approach for spectral dictionary coherence reduction and discuss the feasibility of the methodologies in terms of mutual coherence and approximation error values using overcomplete dictionaries. We compare standard sparse unmixing procedures with our novel derivative method. The presented method was tested on both simulated hyperspectral image as well as on a AVIRIS data.

Selection of numerical measures for pan-sharpening assessment

Abstract: Different tasks of multispectral image analysis and processing require specific properties of input pan-sharpened multispectral data such as spectral and spatial consistency. Generally, the quantitative measures for pan-sharpening assessment were taken from other topics of image processing (e.g. image similarity indexes). All these measures are widely employed for this task but the applicability basis of these measures is not checked and proven. In this paper a comparison of pan-sharpening assessment measures for remote sensing is carried out on specially generated pan-sharpened images. Performed statistical analysis on the assessment measures allows to select the measures which are most sensitive to the pan-sharpened imagery quality and these measures are recommended for use.

Hyperspectral data classification using factor graphs

Abstract: Accurate classification of hyperspectral data is still a competitive task and new classification methods are developed to achieve desired tasks of hyperspectral data use. The objective of this paper is to develop a new method for hyperspectral data classification ensuring the classification model properties like transferability, generalization, probabilistic interpretation, etc. While factor graphs (undirected graphical models) are unfortunately not widely employed in remote sensing tasks, these models possess important properties such as representation of complex systems to model estimation/decision making tasks. In this paper we present a new method for hyperspectral data classification using factor graphs. Factor graph (a bipartite graph consisting of variables and factor vertices) allows factorization of a more complex function leading to definition of variables (employed to store input data), latent variables (allow to bridge abstract class to data), and factors (defining prior probabilities for spectral features and abstract classes; input data mapping to spectral features mixture and further bridging of the mixture to an abstract class). Latent variables play an important role by defining two-level mapping of the input spectral features to a class. Configuration (learning) on training data of the model allows calculating a parameter set for the model to bridge the input data to a class. The classification algorithm is as follows. Spectral bands are separately pre-processed (unsupervised clustering is used) to be defined on a finite domain (alphabet) leading to a representation of the data on multinomial distribution. The represented hyperspectral data is used as input evidence (evidence vector is selected pixelwise) in a configured factor graph and an inference is run resulting in the posterior probability. Variational inference (Mean field) allows to obtain plausible results with a low calculation time. Calculating the posterior probability for each class and comparison of the probabilities leads to classification. Since the factor graphs operate on input data represented on an alphabet (the represented data transferred into multinomial distribution) the number of training samples can be relatively low. Classification assessment on Salinas hyperspectral data benchmark allowed to obtain a competitive accuracy of classification. Employment of training data consisting of 20 randomly selected points for a class allowed to obtain the overall classification accuracy equal to 85.32% and Kappa equal to 0.8358. Representation of input data on a finite domain discards the curse of dimensionality problem allowing to use large hyperspectral data with a moderately high number of bands.

Factor graph models for multisensory data fusion: From low-level features to high level interpretation

Abstract: A solution of difficult tasks in remotely sensed data information extraction can be reached by the development of more complex models. The most important step is in the selection of a relevant and universal methodology for data interpretation, classification, fusion, object detection, etc. Probabilistic graphical models [1] become a more and more popular way for image data annotation and classification [2, 3]. Factor graphs possess important properties such as probabilistic nature, explicit factorization properties, approximate inference, plausible inference of non-full data, easy augmenting, etc., and become relevant for the use in data interpretation systems. In this paper we present several applications of factor graphs for single/multisensory data fusion, classification, and an extension of the graph structure to extract landcover from unseen data. The application of factor graphs allow to obtain an improvement in data fusion/classification accuracy.

Multi-level building reconstruction for automatic enhancement of high resolution dsms

Abstract: In this article a multi-level approach is proposed for reconstruction-based improvement of high resolution Digital Surface Models (DSMs). The concept of Levels of Detail (LOD) defined by CityGML standard has been considered as basis for abstraction levels of building roof structures. Here, the LOD1 and LOD2 which are related to prismatic and parametric roof shapes are reconstructed. Besides proposing a new approach for automatic LOD1 and LOD2 generation from high resolution DSMs, the algorithm contains two generalization levels namely horizontal and vertical. Both generalization levels are applied to prismatic model of buildings. The horizontal generalization allows controlling the approximation level of building footprints which is similar to cartographic generalization concept of the urban maps. In vertical generalization, the prismatic model is formed using an individual building height and continuous to included all flat structures locating in different height levels. The concept of LOD1 generation is based on approximation of the building footprints into rectangular or non-rectangular polygons. For a rectangular building containing one main orientation a method based on Minimum Bounding Rectangle (MBR) in employed. In contrast, a Combined Minimum Bounding Rectangle (CMBR) approach is proposed for regularization of non-rectilinear polygons, i.e. buildings without perpendicular edge directions. Both MBRand CMBR-based approaches are iteratively employed on building segments to reduce the original building footprints to a minimum number of nodes with maximum similarity to original shapes. A model driven approach based on the analysis of the 3D points of DSMs in a 2D projection plane is proposed for LOD2 generation. Accordingly, a building block is divided into smaller parts according to the direction and number of existing ridge lines. The 3D model is derived for each building part and finally, a complete parametric model is formed by merging all the 3D models of the individual parts and adjusting the nodes after the merging step. In order to provide an enhanced DSM, a surface model is provided for each building by interpolation of the internal points of the generated models. All interpolated models are situated on a Digital Terrain Model (DTM) of corresponding area to shape the enhanced DSM. Proposed DSM enhancement approach has been tested on a dataset from Munich central area. The original DSM is created using robust stereo matching of Worldview-2 stereo images. A quantitative assessment of the new DSM by comparing the heights of the ridges and eaves shows a standard deviation of better than 50cm.

Tracing structural changes of a complex forest by a multiple systems approach

Abstract: An investigation on the combination of information from multiple sensors is presented, dealing with the assessment of forest structure and structural changes over time in the complex Mid-European forest of the Traunsteiner Stadtwald, a communal forest in South-East Bavaria, Germany. The starting point of the investigation was a data set from 2003 combining HyMap hyperspectral data and a 0.5m grid DSM calculated from HRSC data. A canopy height model was derived with the help of the official Bavarian State Survey DTM originating from 2001 LIDAR data. During the night of 18 th to 19 th of January 2007 the winter storm Kyrill caused severe damages in that forest. Using satellite data from the systems RapidEye, Cartosat-1 and ALOS Prism the changes in forest structure were analysed. Of special interest was the question whether the parameter derivation accuracy from the lower resolution satellite data are sufficient to assess the damages and to update the data bases of the Traunsteiner Stadtwald forest. The validation of the results was done on behalf of the regular forest inventory data from 1999 and 2009 respectively, supported by a LIDAR data set from 2010 for height assessment of the satellite data derived surface models.

Spectral matching through data compression

Abstract: This paper proposes to use compression-based similarity measures to cluster spectral signatures on the basis of their similarities. Such universal distances estimate the shared information between two objects by comparing their compression factors, which can be obtained by any standard compressor. Experiments on spectra, both collected in the field and selected from a hyperspectral scene, show that these methods may outperform traditional choices for spectral distances based on vector processing such as Spectral Angle, Spectral Information Divergence, Spectral Correlation, and Euclidean Distance.

A novel 3d city modelling approach for satellite stereo data using 3d active shape models on dsms

Abstract: Since remote sensing provides new sensors and techniques to accumulate stereo data on urban regions, three-dimensional (3D) repre- sentation of these regions gained much interest for various applications. 3D urban region representation can e.g. be used for detailed urban monitoring, change and damage detection purposes. In order to obtain 3D representation, one of the easiest and cheapest way is to use Digital Surface Models (DSMs) which are generated from very high resolution stereo satellite images using stereovision tech- niques. Unfortunately after applying the DSM generation process, we cannot directly obtain a full 3D urban region representation. In the DSM which is generated using only one stereo image pair, generally noise, matching errors, and uncertainties on building wall loca- tions are very high. These undesirable effects prevents a DSM to provide a realistic 3D city representation. Therefore, some automatic techniques should be applied to obtain real 3D city models using DSMs as input. In order to bring a solution to the existing problems in this field, herein we propose a fully automated approach based on the usage of a novel 3D active shape model. Our experimental results on DSMs of Munich city which are obtained from different satellite (Cartosat-1, Ikonos, WorldView-2) and airborne sensors (3K camera, HRSC, and LIDAR) indicate possible usage of the algorithm to obtain 3D city representation results automatically.

Information extraction using optical and radar remote sensing data fusion

Abstract: Information extraction from multi-sensor remote sensing imagery is an important and challenging task for many applications such as urban area mapping and change detection. Especially for optical and radar data fusion a special acquisition (orthogonal) geometry is of great importance in order to minimize displacements due to an inaccuracy of the Digital Elevation Model (DEM) used for data ortho-rectification and due to the presence of unknown 3D structures in a scene. Final data spatial alignment is performed manually using ground control points (GCPs) or by a recently proposed automatic co-registration method based on a Mutual Information measure. These data preprocessing steps are of a crucial importance for a success of the following data fusion. For a combination of features originating from different sources, which are quite often non-commensurable, we propose an information fusion framework called INFOFUSE consisting of three main processing steps: feature fission (feature extraction for complete description of a scene), unsupervised clustering (complexity reduction and feature conversion to a common domain) and supervised classification realized by Bayesian/Neural/Graphical networks. Finally, a general data processing chain for multi-sensor data fusion is presented. Examples of buildings in an urban area are presented for very high resolution space borne optical WorldView-2 and radar TerraSAR-X imagery over Munich city, Germany in different acquisition geometries including the orthogonal one. Additionally, theoretical analysis of radar signatures of buildings in urban area and its impact on the joint classification or data fusion is discussed.