Zusammenfassung: Im Beitrag wird die Genauigkeit in der Bestimmung der äußeren Orientierung einer... more Zusammenfassung: Im Beitrag wird die Genauigkeit in der Bestimmung der äußeren Orientierung einer Einzelkamera durch räumlichen Rückwärtsschnitt in Abhängigkeit unterschiedlicher Parameter in einer fiktiven Messumgebung untersucht. Darauf aufbauend wird ein Multi-Kamerasystem betrachtet, das mit vier im 90°-Winkel zueinander montierten Kameras mehrere Raumrichtungen beobachtet. In einem praktischen Versuch mit dem Multi-Kamerasystem werden die realen Abweichungen der äußeren Orientierung bestimmt und mit den theoretisch erreichbaren Genauigkeiten verglichen. Während die Simulation Abweichungen im Bereich unter einem Zehntel Millimeter in der Position liefert, zeigt der praktische Versuch um bis zum Faktor fünf höhere Abweichungen. Der Grund hierfür liegt in der beim praktischen Versuch eingeschränkten Sichtbarkeit von Referenzpunkten.
In dieser Arbeit wird die Oberflächengeschwindigkeit des Taku Gletschers (Juneau, Alaska) mit ein... more In dieser Arbeit wird die Oberflächengeschwindigkeit des Taku Gletschers (Juneau, Alaska) mit einem Korrelationsansatz aus hoch aufgelösten TerraSAR-X Daten mehrerer Überflüge geschätzt. Die Genauigkeit liegt dabei im Bereich von 2 cm/Tag mit einer hohen Zuverlässigkeit. Letztere wird durch einen Vergleich zwischen dem Maximum der Korrelation und seiner gemittelten Nachbarschaft bestimmt. Um die Zuverlässigkeit und die Abdeckung weiter zu erhöhen wird in dieser Arbeit eine Markov Zufallsfeld (MRF) verwendet, um die Daten eines aufsteigenden und abfallenden Orbits zusammenzuführen. Das Verfahren hilft an Stellen, die durch Schatten oder andere Artefakte keine zuverlässigen Messungen erlauben, eine Schätzung der Geschwindigkeit durchzuführen. Des Weiteren wird die Genauigkeit im Gesamten verbessert, da redundante Informationen konsequent genutzt und Widersprüche durch das Modell aufgelöst werden.
This article presents an investigation on the co-registration of a 3d point cloud and intensity i... more This article presents an investigation on the co-registration of a 3d point cloud and intensity image of a near-infrared (NIR) time-of-flight (TOF) camera system and a thermal infrared (TIR) camera. The TIR-camera is relatively oriented to the TOF-camera. As the radiometric behavior of the TIR-camera and the NIR-image of the TOF-camera is different for most pixels, corners are detected in both images and co-registered. The intensity values of TIR-camera are projected into the TOF point cloud and from there in the TOF-camera image plane. The quality of the projection is evaluated comparing the projected TIR image from the view of the TOF-camera and the TOF NIR image.
This paper shows the potential of airborne single-pass millimeterwave InSAR data for the recognit... more This paper shows the potential of airborne single-pass millimeterwave InSAR data for the recognition of individual trees. In addition to the very high available resolution in the decimeter range, the benefits of both single-pass interferometry as well as millimeterwave radar are discussed. Using knowledge of the side-looking imaging geometry, the representation of trees in airborne SAR data is modeled and compared to optical and LiDAR remote sensing data. Finally, two different concepts for automated single tree recognition are proposed in order to show the versatile potential of synthetic aperture radar interferometry for the required task.
In dieser Arbeit wird ein Konzept zur Klassifikation urbaner Strukturen in SAR Intensitätsbildern... more In dieser Arbeit wird ein Konzept zur Klassifikation urbaner Strukturen in SAR Intensitätsbildern vorgestellt. Die Methode basiert dabei auf der Verwendung von Conditional Random Fields (CRF), deren Netzwerkstruktur und Parametrisierung automatisch durch das Optimieren einer Zielfunktion gelernt werden soll. Allerdings handelt es sich um ein schlecht gestelltes Problem, was deutliche Herausforderungen an die Algorithmen stellt. Wir zeigen einen effektiven Weg, um dennoch eine Optimierung erreichen zu können und welchen Vorteil regelbasierte Verfahren, die durch den Nutzer definieret werden, in diesem Zusammenhang haben.
The increasing resolution of modern SAR sensors leads to new approaches for object detection in S... more The increasing resolution of modern SAR sensors leads to new approaches for object detection in SAR images. Objects appear not just as blobs but rather show a certain structure which can be exploited by advanced approaches already established in image processing. But an increase in detail comes at the cost of an increased complexity. In this work we use Mathematical Morphology (MM) to detect stationary vehicles in airborne decimeter resolution SAR intensity images. Mathematical Morphology has become popular because of its proven utility and sound mathematical foundations. Here we use the concept of connected components to simplify the image and to describe parts of the vehicles. The detection is a subsequent process. It is shown that a robust detection based on the explicit description of the vehicle appearance is possible.
ABSTRACT: The analysis of airborne laser scanner data to extract surface features is of great int... more ABSTRACT: The analysis of airborne laser scanner data to extract surface features is of great interest in photogrammetric research. Especially for applications based on airborne measurements, where the intensity is crucial (eg for segmentation, classification or visualization purposes), a normalization considering the beam divergence, the incidence angle and the atmospheric attenuation is required.
ABSTRACT: In this paper we describe investigations for digital recording of received laser pulses... more ABSTRACT: In this paper we describe investigations for digital recording of received laser pulses and a detailed analysis of the pulse waveform. An experimental system was constructed using an airborne laser scanning system and a receiver unit with a high-resolution sampling rate, which is above the sampling rate of current commercial laser systems. Capturing the temporal waveform by simultaneous scanning of the scene delivers a data volume (space-time cube).
ABSTRACT: The estimation of the transformation parameters between different point clouds is still... more ABSTRACT: The estimation of the transformation parameters between different point clouds is still a crucial task as it is usually followed by scene reconstruction, object detection or object recognition. Therefore, the estimates should be as accurate as possible. Recent developments show that it is feasible to utilize both the measured range information and the reflectance information sampled as image, as 2D imagery provides additional information.
Abstract: The real and imaginary parts are proposed as an alternative to the usual Polar represen... more Abstract: The real and imaginary parts are proposed as an alternative to the usual Polar representation of complex-valued images. It is proven that the transformation from Polar to Cartesian representation contributes to decreased mutual information, and hence to greater distinctiveness. The Complex Scale-Invariant Feature Transform (ℂSIFT) detects distinctive features in complex-valued images.
ABSTRACT: The new sensor technique for receiving full-waveform laser data delivers not only heigh... more ABSTRACT: The new sensor technique for receiving full-waveform laser data delivers not only height information but also additional features like intensity. The echoes generated by reflecting objects are approximated by amplitude and width of a Gaussian kernel. These features are used for tree region segmentation. The inclusion of the available information in the environment around each 3d point of the cloud can be used to stabilize the segmentation process.
ABSTRACT: For the first time the basic principle to unify advantages between active sensors and t... more ABSTRACT: For the first time the basic principle to unify advantages between active sensors and the simultaneous capturing of an image for an extended area of dynamical 3D applications in close range is given by range imaging (RIM) sensors. The drawback of data which is captured with RIM sensors is the absolute range accuracy and the limited non-ambiguity range.
Abstract: The analysis of laser data to extract surface features is of great interest. Beside the... more Abstract: The analysis of laser data to extract surface features is of great interest. Beside the range values, various information about features of the illuminated surface can be gained. In this paper, the surface features range, roughness, and reflectance are determined using different algorithms for extracting pulse properties. The accuracy of the received results is investigated by the comparison of two measurements of the same scene at different points in time to evaluate the robustness of feature extraction.
Abstract The analysis of data derived by full-waveform laser scanning systems is of great interes... more Abstract The analysis of data derived by full-waveform laser scanning systems is of great interest. In this study, we use a simulated surface response to estimate the slope of a plane surface by full-waveform analysis. For analysis the transmitted waveform of the emitted pulse is used to estimate the received waveform of the backscattered pulse for a known surface. We simulated a plane surface with different slopes. Typical spatial beam distributions are considered for modeling, namely Gaussian and uniform.
ABSTRACT: Obtaining a 3D description of man-made and natural environments is a basic task in Comp... more ABSTRACT: Obtaining a 3D description of man-made and natural environments is a basic task in Computer Vision and Remote Sensing. To this end, laser scanning is currently one of the dominating techniques to gather reliable 3D information. The scanning principle inherently needs a certain time interval to acquire the 3D point cloud. On the other hand, new active sensors provide the possibility of capturing range information by images with a single measurement.
ABSTRACT: Laser range data analysis is of high interest in photogrammetry. Range estimation for c... more ABSTRACT: Laser range data analysis is of high interest in photogrammetry. Range estimation for complex surface structures can be inaccurate. To overcome this drawback a method using a laser scanner capable of full-waveform analysis is proposed. For analysis the transmitted waveform of the emitted pulse is used to estimate the received waveform of the backscattered pulse for a known surface. We simulated a plane surface with different slopes and a sphere.
ABSTRACT: In this paper we describe a simulation system and investigations for analysis of record... more ABSTRACT: In this paper we describe a simulation system and investigations for analysis of recorded laser pulses. A simulation setup that considers simulated signals of synthetic objects was developed for exploring the capabilities of recognizing urban objects using a laser system. The temporal waveform of each laser pulse is analyzed for gaining the pulse properties: range, pulse power and number of peaks.
Kurzfassung Bei gepulsten Lasersystemen erfolgt die Bestimmung der Entfernung zu einem ange% leuc... more Kurzfassung Bei gepulsten Lasersystemen erfolgt die Bestimmung der Entfernung zu einem ange% leuchteten Objekt durch Messung der Laufzeit eines Pulses. Die Zeitmessung wird ty% pischerweise aus einem charakteristischen Wert der empfangenen Signalform des rück% gestreuten Pulses bestimmt. Kritische Messsituationen ergeben sich, falls die zeitliche Signalform des rückgestreuten Pulses deformiert ist oder mehr als ein Puls rückgestreut wird.
ABSTRACT: Laser range data is of high interest in photogrammetry. However, compared to passive im... more ABSTRACT: Laser range data is of high interest in photogrammetry. However, compared to passive imagery, it usually has a lower image resolution, making geometry extraction and modeling challenging. A method to overcome this handicap using a laser scanner capable of full waveform analysis is proposed. The recorded pulse waveform is analyzed to find range and intensity values for all of the surfaces in the beam footprint.
ABSTRACT: The analysis of data derived by full-waveform laser scanning systems is of great intere... more ABSTRACT: The analysis of data derived by full-waveform laser scanning systems is of great interest. In this study, we estimate the impulse response of a laser scanning system capturing the waveform. Considering the impulse response of the system is important for precise waveform analysis. The characteristic of the system has to be mentioned to interpret the pulse properties in an accurate way.
Zusammenfassung: Im Beitrag wird die Genauigkeit in der Bestimmung der äußeren Orientierung einer... more Zusammenfassung: Im Beitrag wird die Genauigkeit in der Bestimmung der äußeren Orientierung einer Einzelkamera durch räumlichen Rückwärtsschnitt in Abhängigkeit unterschiedlicher Parameter in einer fiktiven Messumgebung untersucht. Darauf aufbauend wird ein Multi-Kamerasystem betrachtet, das mit vier im 90°-Winkel zueinander montierten Kameras mehrere Raumrichtungen beobachtet. In einem praktischen Versuch mit dem Multi-Kamerasystem werden die realen Abweichungen der äußeren Orientierung bestimmt und mit den theoretisch erreichbaren Genauigkeiten verglichen. Während die Simulation Abweichungen im Bereich unter einem Zehntel Millimeter in der Position liefert, zeigt der praktische Versuch um bis zum Faktor fünf höhere Abweichungen. Der Grund hierfür liegt in der beim praktischen Versuch eingeschränkten Sichtbarkeit von Referenzpunkten.
In dieser Arbeit wird die Oberflächengeschwindigkeit des Taku Gletschers (Juneau, Alaska) mit ein... more In dieser Arbeit wird die Oberflächengeschwindigkeit des Taku Gletschers (Juneau, Alaska) mit einem Korrelationsansatz aus hoch aufgelösten TerraSAR-X Daten mehrerer Überflüge geschätzt. Die Genauigkeit liegt dabei im Bereich von 2 cm/Tag mit einer hohen Zuverlässigkeit. Letztere wird durch einen Vergleich zwischen dem Maximum der Korrelation und seiner gemittelten Nachbarschaft bestimmt. Um die Zuverlässigkeit und die Abdeckung weiter zu erhöhen wird in dieser Arbeit eine Markov Zufallsfeld (MRF) verwendet, um die Daten eines aufsteigenden und abfallenden Orbits zusammenzuführen. Das Verfahren hilft an Stellen, die durch Schatten oder andere Artefakte keine zuverlässigen Messungen erlauben, eine Schätzung der Geschwindigkeit durchzuführen. Des Weiteren wird die Genauigkeit im Gesamten verbessert, da redundante Informationen konsequent genutzt und Widersprüche durch das Modell aufgelöst werden.
This article presents an investigation on the co-registration of a 3d point cloud and intensity i... more This article presents an investigation on the co-registration of a 3d point cloud and intensity image of a near-infrared (NIR) time-of-flight (TOF) camera system and a thermal infrared (TIR) camera. The TIR-camera is relatively oriented to the TOF-camera. As the radiometric behavior of the TIR-camera and the NIR-image of the TOF-camera is different for most pixels, corners are detected in both images and co-registered. The intensity values of TIR-camera are projected into the TOF point cloud and from there in the TOF-camera image plane. The quality of the projection is evaluated comparing the projected TIR image from the view of the TOF-camera and the TOF NIR image.
This paper shows the potential of airborne single-pass millimeterwave InSAR data for the recognit... more This paper shows the potential of airborne single-pass millimeterwave InSAR data for the recognition of individual trees. In addition to the very high available resolution in the decimeter range, the benefits of both single-pass interferometry as well as millimeterwave radar are discussed. Using knowledge of the side-looking imaging geometry, the representation of trees in airborne SAR data is modeled and compared to optical and LiDAR remote sensing data. Finally, two different concepts for automated single tree recognition are proposed in order to show the versatile potential of synthetic aperture radar interferometry for the required task.
In dieser Arbeit wird ein Konzept zur Klassifikation urbaner Strukturen in SAR Intensitätsbildern... more In dieser Arbeit wird ein Konzept zur Klassifikation urbaner Strukturen in SAR Intensitätsbildern vorgestellt. Die Methode basiert dabei auf der Verwendung von Conditional Random Fields (CRF), deren Netzwerkstruktur und Parametrisierung automatisch durch das Optimieren einer Zielfunktion gelernt werden soll. Allerdings handelt es sich um ein schlecht gestelltes Problem, was deutliche Herausforderungen an die Algorithmen stellt. Wir zeigen einen effektiven Weg, um dennoch eine Optimierung erreichen zu können und welchen Vorteil regelbasierte Verfahren, die durch den Nutzer definieret werden, in diesem Zusammenhang haben.
The increasing resolution of modern SAR sensors leads to new approaches for object detection in S... more The increasing resolution of modern SAR sensors leads to new approaches for object detection in SAR images. Objects appear not just as blobs but rather show a certain structure which can be exploited by advanced approaches already established in image processing. But an increase in detail comes at the cost of an increased complexity. In this work we use Mathematical Morphology (MM) to detect stationary vehicles in airborne decimeter resolution SAR intensity images. Mathematical Morphology has become popular because of its proven utility and sound mathematical foundations. Here we use the concept of connected components to simplify the image and to describe parts of the vehicles. The detection is a subsequent process. It is shown that a robust detection based on the explicit description of the vehicle appearance is possible.
ABSTRACT: The analysis of airborne laser scanner data to extract surface features is of great int... more ABSTRACT: The analysis of airborne laser scanner data to extract surface features is of great interest in photogrammetric research. Especially for applications based on airborne measurements, where the intensity is crucial (eg for segmentation, classification or visualization purposes), a normalization considering the beam divergence, the incidence angle and the atmospheric attenuation is required.
ABSTRACT: In this paper we describe investigations for digital recording of received laser pulses... more ABSTRACT: In this paper we describe investigations for digital recording of received laser pulses and a detailed analysis of the pulse waveform. An experimental system was constructed using an airborne laser scanning system and a receiver unit with a high-resolution sampling rate, which is above the sampling rate of current commercial laser systems. Capturing the temporal waveform by simultaneous scanning of the scene delivers a data volume (space-time cube).
ABSTRACT: The estimation of the transformation parameters between different point clouds is still... more ABSTRACT: The estimation of the transformation parameters between different point clouds is still a crucial task as it is usually followed by scene reconstruction, object detection or object recognition. Therefore, the estimates should be as accurate as possible. Recent developments show that it is feasible to utilize both the measured range information and the reflectance information sampled as image, as 2D imagery provides additional information.
Abstract: The real and imaginary parts are proposed as an alternative to the usual Polar represen... more Abstract: The real and imaginary parts are proposed as an alternative to the usual Polar representation of complex-valued images. It is proven that the transformation from Polar to Cartesian representation contributes to decreased mutual information, and hence to greater distinctiveness. The Complex Scale-Invariant Feature Transform (ℂSIFT) detects distinctive features in complex-valued images.
ABSTRACT: The new sensor technique for receiving full-waveform laser data delivers not only heigh... more ABSTRACT: The new sensor technique for receiving full-waveform laser data delivers not only height information but also additional features like intensity. The echoes generated by reflecting objects are approximated by amplitude and width of a Gaussian kernel. These features are used for tree region segmentation. The inclusion of the available information in the environment around each 3d point of the cloud can be used to stabilize the segmentation process.
ABSTRACT: For the first time the basic principle to unify advantages between active sensors and t... more ABSTRACT: For the first time the basic principle to unify advantages between active sensors and the simultaneous capturing of an image for an extended area of dynamical 3D applications in close range is given by range imaging (RIM) sensors. The drawback of data which is captured with RIM sensors is the absolute range accuracy and the limited non-ambiguity range.
Abstract: The analysis of laser data to extract surface features is of great interest. Beside the... more Abstract: The analysis of laser data to extract surface features is of great interest. Beside the range values, various information about features of the illuminated surface can be gained. In this paper, the surface features range, roughness, and reflectance are determined using different algorithms for extracting pulse properties. The accuracy of the received results is investigated by the comparison of two measurements of the same scene at different points in time to evaluate the robustness of feature extraction.
Abstract The analysis of data derived by full-waveform laser scanning systems is of great interes... more Abstract The analysis of data derived by full-waveform laser scanning systems is of great interest. In this study, we use a simulated surface response to estimate the slope of a plane surface by full-waveform analysis. For analysis the transmitted waveform of the emitted pulse is used to estimate the received waveform of the backscattered pulse for a known surface. We simulated a plane surface with different slopes. Typical spatial beam distributions are considered for modeling, namely Gaussian and uniform.
ABSTRACT: Obtaining a 3D description of man-made and natural environments is a basic task in Comp... more ABSTRACT: Obtaining a 3D description of man-made and natural environments is a basic task in Computer Vision and Remote Sensing. To this end, laser scanning is currently one of the dominating techniques to gather reliable 3D information. The scanning principle inherently needs a certain time interval to acquire the 3D point cloud. On the other hand, new active sensors provide the possibility of capturing range information by images with a single measurement.
ABSTRACT: Laser range data analysis is of high interest in photogrammetry. Range estimation for c... more ABSTRACT: Laser range data analysis is of high interest in photogrammetry. Range estimation for complex surface structures can be inaccurate. To overcome this drawback a method using a laser scanner capable of full-waveform analysis is proposed. For analysis the transmitted waveform of the emitted pulse is used to estimate the received waveform of the backscattered pulse for a known surface. We simulated a plane surface with different slopes and a sphere.
ABSTRACT: In this paper we describe a simulation system and investigations for analysis of record... more ABSTRACT: In this paper we describe a simulation system and investigations for analysis of recorded laser pulses. A simulation setup that considers simulated signals of synthetic objects was developed for exploring the capabilities of recognizing urban objects using a laser system. The temporal waveform of each laser pulse is analyzed for gaining the pulse properties: range, pulse power and number of peaks.
Kurzfassung Bei gepulsten Lasersystemen erfolgt die Bestimmung der Entfernung zu einem ange% leuc... more Kurzfassung Bei gepulsten Lasersystemen erfolgt die Bestimmung der Entfernung zu einem ange% leuchteten Objekt durch Messung der Laufzeit eines Pulses. Die Zeitmessung wird ty% pischerweise aus einem charakteristischen Wert der empfangenen Signalform des rück% gestreuten Pulses bestimmt. Kritische Messsituationen ergeben sich, falls die zeitliche Signalform des rückgestreuten Pulses deformiert ist oder mehr als ein Puls rückgestreut wird.
ABSTRACT: Laser range data is of high interest in photogrammetry. However, compared to passive im... more ABSTRACT: Laser range data is of high interest in photogrammetry. However, compared to passive imagery, it usually has a lower image resolution, making geometry extraction and modeling challenging. A method to overcome this handicap using a laser scanner capable of full waveform analysis is proposed. The recorded pulse waveform is analyzed to find range and intensity values for all of the surfaces in the beam footprint.
ABSTRACT: The analysis of data derived by full-waveform laser scanning systems is of great intere... more ABSTRACT: The analysis of data derived by full-waveform laser scanning systems is of great interest. In this study, we estimate the impulse response of a laser scanning system capturing the waveform. Considering the impulse response of the system is important for precise waveform analysis. The characteristic of the system has to be mentioned to interpret the pulse properties in an accurate way.
Uploads
Papers by Uwe Stilla