Leonie S. Dreschler-Fischer

Lernen durch Beobachtung von Szenen mit bewegten Objekten: Phasen einer Systementwicklung

Das MORIO-System (MORIO = MOving RIgid Object description) dient zur Analyse von monokularen schwarz/weis TV-Bildfolgen, die mit einer stationaren Kamera aufgezeichnet wurden. Die Aufgabe dieses Systems besteht darin, dreidimensionale Modelle aus mehrfachen Ansichten eines bewegten starren Korpers zu lernen, ohne szenenspezifisches Vorwissen uber den Inhalt der Bildfolge zu verwenden. Als Ergebnis wird ein approximatives volumetrisches Modell berechnet, und zwar die konvexe Hulle der rekonstruierten Objektpunkte. Dieses Modell kann als Ausgangspunkt zur genaueren Modellierung der Objektoberflachen verwendet werden [WESTPHAL 82].

Konzeption einer virtuellen Maschine als Standardschnittstelle für die Bildverarbeitung

Die erste Generation von Darstellungsgeraten in der Bildverarbeitung verfugte uber keine oder nur minimale Verarbeitungsmoglichkeiten, die beispielsweise durch Funktionstabellen realisiert waren. Mit der Zeit kristallisierten sich jedoch Operationen heraus, die in vielen Anwendungsbereichen eingesetzt werden und daher als Standardoperationen angesehen werden konnen. Gesunkene Hardwarekosten, verbunden mit der Erkenntnis, das sich viele komplexe Operationen aus elementaren Standardoperationen konstruieren lassen, fuhrten zur Entwicklung von Bildverarbeitungsgeraten, die einfache Operationen, wie den Laplace-Operator, auf den in ihnen gespeicherten Bildern ausfuhren konnen. Im Zuge dieser Entwicklung entstand eine Vielzahl von Geraten, die sich bezuglich ihres Funktionsumfangs und ihrer Benutzeroberflache stark unterscheiden.

Two-dimensional sea surface current fields derived from multi-sensor satellite data

We used multi-sensor satellite images to compute small-scale surface currents in the Northern Baltic Sea. The images were acquired by the Thematic Mapper (TM) aboard Landsat, by the Synthetic Aperture Radar (SAR) aboard ERS-2, and by the Wide-Field Scanner (WiFS) aboard IRS-1C within two hours on 15 July, 1997, over the Northern Baltic Proper. Most of them show manifestations of an ongoing algae bloom, which are driven by the local currents. Pairs of images acquired at either different electromagnetic frequencies (TM and SAR) or at the same electromagnetic bands (TM and WiFS) are used for applying cross-correlation or partial differentiation techniques, respectively. Our results are insofar promising as the computed two-dimensional surface current fields complement existing data from numerical models. However, limitations of the proposed method are due to the strong dependence of the visibility of algae bloom manifestations on local weather conditions and to the low availability of satellite data.

Detecting and tracking small scale eddies in the black sea and the Baltic Sea using high-resolution Radarsat-2 and TerraSAR-X imagery (DTeddie)

High-resolution Synthetic Aperture radar (SAR) images of coastal and off-coastal areas provide a nearly weather independent monitoring of small-scale oceanographic features. In this paper, we describe the advances, which have been achieved within the DTeddie project. Therefore, we present the application of formerly developed algorithms to high-resolution SAR data to estimate the sea surface current at sub-mesoscale. Inside the derived current sea surface current fields, vortical structures, e.g. eddies, will be detected by means of local current anomalies. Another approach uses the alignment of anchored ships as an indicator of the sea surface current. To infer dependent information for each ship, we present an automatic registration method between coastal photographs and the SAR images. Both approaches have proven to yield valuable results with respect to the detection of vortical current flows at sub-mesoscale.

Fast Derivation of Soil Surface Roughness Parameters Using Multi-band SAR Imagery and the Integral Equation Model

The Integral Equation Model (IEM) predicts the normalized radar cross section (NRCS) of dielectric surfaces given surface and radar parameters. To derive the surface parameters from the NRCS using the IEM, the model needs to be inverted. We present a fast method of this model inversion to derive soil surface roughness parameters from synthetic aperture radar (SAR) remote sensing data. The model inversion is based on two different collocated SAR images of different bands, the derivation of the parameters cannot be done using one band alone. The computation of the model and the model inversion are very time consuming tasks and therefore may be impractical for large remote sensing data. We present an approach that is based on a few model assumptions to speed up the computation of the surface parameters. We applied the algorithm to detect the correlation length of the surface for dry-fallen areas in the World Cultural Heritage ”Wadden Sea”, a coastal tidal flat at the German Bight (North Sea). The results are very promising and may be used for a classification of the area in future steps.

Using Snakes with Asymmetric Energy Terms for the Detection of Varying-Contrast Edges in SAR Images

Active contour methods like snakes, have become a basic tool in computer vision and image analysis over the last years. They have proven to be adequate for the task of finding boundary features like broken edges in an image. However, when applying the basic snake technique to synthetic aperture radar (SAR) remote sensing images, the detection of varying-contrast edges may not be satisfying. This is caused by the special imaging technique of SAR and the commonly known specklenoise. In this paper we propose the use of asymmetric external energy terms to cope with this problem. We show first results of the method for the detection of edges of tidal creeks using an ENVISAT ASAR image. These creeks can be found in the World Heritage Site ”Wadden Sea” located at the German Bight (North Sea).

Ein geometrisches Modell für die Zellsegmentierung

Die exakte Segmentierung von Zellen in Mikroskopaufnahmen gehort seit jeher zu den aktiven Forschungsgebieten der Bildverarbeitung. Herkommliche Verfahren setzen allerdings hochstens indirekt auf eine genaue Modellierung der Zellmorphologie. Im vorliegenden Beitrag wird ein geometrisches Modell aufgestellt, das diesem Mangel begegnet.

A knowledge-based approach to the detection and interpretation of changes in aerial images

Traditional approaches to change detection do not give information about the cause of changes and, therefore, do not help to automatically evaluate their significance and relevance with respect to a given analysis task. In this contribution the authors present a knowledge-based approach to the problem of interpreting changes which allows one to compare information on a symbolic image description level rather than on a pure data level. The authors introduce a system architecture and discuss the structure and function of its main three components.<<ETX>>

From multi-sensor tracking of sea surface films to mesoscale and sub-mesoscale sea surface current fields

The knowledge about mesoscale and sub-mesoscale sea surface current fields is of high interdisciplinary interest, since it results in a better understanding of ocean-atmosphere interactions. However, many available numerical model results are of resolutions, which are too coarse to investigate mesoscale and sub-mesoscale turbulent features like eddies, particularly in coastal waters. In this work we present the results of tracking biogenic and anthropogenic surface film signatures on multi-sensor satellite images (SAR and multispectral images) to estimate the local sea surface current field. The main advantage of this approach is that the resolution of the derived current fields depends mainly on the resolution of the images, which have been used for tracking. Due to the large temporal distance between two acquisitions of a scene and the high variability of the tracked sea surface films, classical tracking methods, e.g. feature based or Optical Flow methods may not be applicable to successfully track the imaged signatures of the surface films. In this work, we use our former developed generic framework, which ensures the applicability and increases the stability of the results for wellknown tracking and Optical Flow algorithms. With this framework, it is e.g. possible to compute the sea surface current field using Optical Flow approaches even for large spatiotemporal distances and with partial scene coverage. We present and compare the results of different tracking algorithms by means of tracking biogenic sea surface films. The investigated areas are the Baltic Sea and the Black Sea. We present the use of Landsat TM, SeaWiFS, ERS-2, TerraSARX and RADARSAT-2 data for the derivation of sea surface currents. The resolution of the images used varies from moderate to fine resolution, which allows the derivation sea surface current fields of moderate to fine resolutions.

A multispectral edge detection scheme and its application to Landsat imagery

It is shown how to extend in a formally well-founded way a state-of-the-art gray-level edge detector to be applied to multi-dimensional image functions. Based on the mathematical approach the authors introduce a detection scheme which is not limited to a specific number of image components (spectral channels). In several experiments done on TM-data The authors demonstrate that the multispectral approach allows them to integrate the contrast information contained in the different channels in a satisfactory way.<<ETX>>