Bio: Eckart Michaelsen is an academic researcher from Fraunhofer Society. The author has contributed to research in topics: Gestalt psychology & Pattern recognition (psychology). The author has an hindex of 14, co-authored 75 publications receiving 703 citations.
Papers published on a yearly basis
01 Jan 2004
TL;DR: Three major frequency domains of remote sensing are considered, namely (i) visual, (ii) thermal IR and (iii) radar, which allows remote sensing on a day-night basis and under bad weather conditions.
Abstract: In this paper several possibilities of vehicle extraction from different airborne sensor systems are described. Three major frequency domains of remote sensing are considered, namely (i) visual, (ii) thermal IR and (iii) radar. Due to the complementing acquired scene properties, the image processing methods have to be tailored for the peculiarities of the different kinds of sensor data. (i)Vehicle detection in aerial images relies upon local and global features. For modelling a vehicle on the local level, a 3Dwireframe representation is used describing prominent geometric and radiometric features of cars including their shadow region. A vehicle is extracted by a ”top-down” matching of the model to the image. On the global level, vehicle queues are modelled by ribbons that exhibit typical symmetries and spacing of vehicles over a larger distance. Fusing local and global extractions makes the result more complete. (ii) Particularly at night video sequences from an infrared camera yield suitable data to assess the activity of vehicles. At the resolution of approximately one meter vehicles appear as elongated spots. However, in urban areas many additional other objects have the same property. Vehicles may be discriminated from these objects either by their movement or by their temperature and their appearance in groups. Using map knowledge as context, a grouping of vehicles into rows along road margins is performed. (iii) The active scene illumination and large signal wavelength of SAR allows remote sensing on a day-night basis and under bad weather conditions. High-resolution SAR systems open the possibility to detect objects like vehicles and to determine the velocity of moving objects. Along-track interferometry allows estimation even small vehicle movements. However, in urban areas SAR specific illumination phenomena like foreshortening, layover, shadow, and multipath-propagation burden the interpretation. Particularly the visibility of the vehicles in inner city areas is in question. A high resolution LIDAR DEM is incorporated to determine the visibility of the roads by a SAR measurement from a given sensor trajectory and sensor orientation. Shadow and layover areas are detected by incoherent sampling of the DEM. In order to determine the optimal flight path a large number of simulations are carried out with varying viewing and aspect angles. * Corresponding author (firstname.lastname@example.org).
TL;DR: In this paper, a knowledge-based approach is applied, which is realized by a production system that codes a set of suitable principles of perceptual grouping in its production rules, which combines and matches these 2D image objects and infer their height by 3D clustering.
Abstract: SAR stereo image analysis for 3D information extraction is mostly carried out based on imagery taken under same-side or opposite-side viewing conditions. For urban scenes in practice stereo is up to now usually restricted to the first configuration, because increasing image dissimilarity connected with rising illumination direction differences leads to a lack of suitable features for matching, especially in the case of low or medium resolution data. However, due to two developments SAR stereo from arbitrary viewing conditions becomes an interesting option for urban information extraction. The first one is the availability of airborne sensor systems, which are capable of more flexible data acquisition in comparison to satellite sensors. This flexibility enables multi-aspect analysis of objects in built-up areas for various kinds of purpose, such as building recognition, road network extraction, or traffic monitoring. The second development is the significant improvement of the geometric resolution providing a high level of detail especially of roof features, which can be observed from a wide span of viewpoints. In this paper, high-resolution SAR images of an urban scene are analyzed in order to infer buildings and their height from the different layover effects in views taken from orthogonal aspect angles. High level object matching is proposed that relies on symbolic data, representing suitable features of urban objects. Here, a knowledge-based approach is applied, which is realized by a production system that codes a set of suitable principles of perceptual grouping in its production rules. The images are analyzed separately for the presence of certain object groups and their characteristics frequently appearing on buildings, such as salient rows of point targets, rectangular structures or symmetries. The stereo analysis is then accomplished by means of productions that combine and match these 2D image objects and infer their height by 3D clustering. The approach is tested using real SAR data of an urban scene.
TL;DR: Extended building features such as long thin roof edge lines, groups of salient point scatterers, and symmetric configurations are detected using principles from perceptual grouping and Gestalt psychology, which are good continuation, similarity, proximity and symmetry.
01 Jan 1997
TL;DR: A representational scheme for the analysis of man-made structures in aerial images and maps is described and two example nets are given to demonstrate the flexibility and applicability of the approach.
Abstract: A representational scheme for the analysis of man-made structures in aerial images and maps is described. Knowledge about object structures is represented by a set of productions. The interaction of the productions is depicted by production nets. The approach is discussed in relation to similar representations. Two example nets are given to demonstrate the flexibility and applicability of the approach. The first one is on the automatic 3D structure analysis of suburban scenes in series of aerial images. The second is on the automatic construction of descriptions of complex buildings in vector maps.
01 Jan 2006
TL;DR: The comparison is performed on synthetic and real data and is based on standard statistical methods, where GOODSAC achieves higher precision than RANSAC.
Abstract: GOODSAC is a paradigm for estimation of model parameters given measurements that are contaminated by outliers. Thus, it is an alternative to the well known RANSAC strategy. GOODSAC’s search for a proper set of inliers does not only maximize the sheer size of this set, but also takes other assessments for the utility into account. Assessments can be used on many levels of the process to control the search and foster precision and proper utilization of the computational resources. This contribution discusses and compares the two methods. In particular, the estimation of essential matrices is used as example. The comparison is performed on synthetic and real data and is based on standard statistical methods, where GOODSAC achieves higher precision than RANSAC.
03 Jun 2014
TL;DR: The Navigation and Control technology embedded in a recently commercialized micro Unmanned Aerial Vehicle (UAV), the AR.Drone, relies on state-of-the-art indoor navigation systems combining low-cost inertial sensors, computer vision techniques, sonar, and accounting for aerodynamics models.
TL;DR: A new database of aerial images provided as a tool to benchmark automatic target recognition algorithms in unconstrained environments and gives the performance of baseline algorithms on this dataset, for different settings of these algorithms, to illustrate the difficulties of the task and provide baseline comparisons.
TL;DR: This paper surveys the state-of-the-art automatic object extraction techniques from aerial imagery and focuses on building extraction approaches, which present the majority of the work in this area.
TL;DR: In spite of many remaining unsolved problems and need for further research and development, use of knowledge and semi-automation are the only viable alternatives towards development of useful object extraction systems, as some commercial systems on building extraction and 3D city modelling as well as advanced, practically oriented research have shown.
Abstract: The paper focuses mainly on extraction of important topographic objects, like buildings and roads, that have received much attention the last decade. As main input data, aerial imagery is considered, although other data, like from laser scanner, SAR and high-resolution satellite imagery, can be also used. After a short review of recent image analysis trends, and strategy and overall system aspects of knowledge-based image analysis, the paper focuses on aspects of knowledge that can be used for object extraction: types of knowledge, problems in using existing knowledge, knowledge representation and management, current and possible use of knowledge, upgrading and augmenting of knowledge. Finally, an overview on commercial systems regarding automated object extraction and use of a priori knowledge is given. In spite of many remaining unsolved problems and need for further research and development, use of knowledge and semi-automation are the only viable alternatives towards development of useful object extraction systems, as some commercial systems on building extraction and 3D city modelling as well as advanced, practically oriented research have shown.