Current page: Members->List of Members->Heinrich Niemann
 

Heinrich Niemann
Universität Erlangen-Nürnberg
Membership Number: 17
Address: Institut für Informatik, Martensstr. 3, D-91058, Erlangen, Germany
Email: niemann@informatik.uni-erlangen.de
Phone: +49 9131 85-27775
Fax: +49 9131 303811
URL: http://www.informatik.uni-erlangen.de

Biographical Sketch
Heinrich Niemann obtained the degree of Dipl.-Ing. In Electrical Engineering and Dr.-Ing. at the Technical University Hannover in 1966 and 1969, respectively. During 1966/67, was a graduate student at the University of Illinois, Urbana. From, 1967 to 1972, he was with the Faunhofer Institut für Informationsverarbeitung in Technik und Biologie, Karlsruhe, working in the field of pattern recognition and biological cybernetics. During 1973-1975, he was teaching at Fachhochschule Giessen in the Department of Electrical Engineering. Since 1975, he has been Professor of Computer Science at the University of Erlangen-Nürnberg, where he was dean of the engineering faculty of the university from 1979-1981. Since 1988 he is also head of the research group ‘Knowledge Processing’ at the Bavarian Research Institute for Knowledge Based Systems (FORWISS), and he was on the board of directors for six years. Since 1998, he is also speaker of a ‘special research area’ (SFB) entitled ‘Model-Based Analysis and Visualization of Complex Scenes and Sensor Data’ funded by the German Research Foundation (DFG).

His fields of research are speech and image understanding and the application of artificial intelligence techniques in these fields. He is on the editorial board of Signal Processing, Pattern Recognition Letters, Pattern Recognition and Image Analysis, Journal of Computing and Information Technology, and Computers and Electrical Engineering. He is the author or coauthor of seven books and about 400 journal and conference contributions as well as editor or coeditor of 24 proceedings, volumes and special issues. He is a member of DAGM, ESCA, EURASIP, GI, IEEE, and VDE.

University of Erlangen (Institute for Pattern Recognition)
The Institute for Pattern Recognition from the University of Erlangen has been working for 20 years in the area of image processing and analysis. Currently research is done in 3D object recognition, 3D object tracking, optimal sensor data acquisition and fusion, camera calibration, 3D reconstruction, image-based object and scene modelling, augmented reality, and medical image processing:

- Optimal camera parameter selection is one current research goal in several applications. This includes optimal viewpoint selection for object localization and recognition, gaze control for robot self-localization and optimal focal length selection in object tracking. Extensions in reinforcement learning and an information theoretic framework have been developed to select optimal sensor data to solve a given problem. Closely related is the problem of sensor data fusion for which particle filter approaches are exploited.

- Appearance-based 3D object recognition is done using statistical and eigenspace approaches as well as neural networks. For the statistical methods, object models are represented by mixture densities of object appearance or features in the eigenspace, with respect to camera geometry, lighting, colour, and object size.

- Structure from motion approaches have been applied for reconstruction of plenoptic scene models (lightfields) from sequences of hand-held cameras. The new technique serves as an alternative to classical geometric descriptions of object scenes. In computer vision plenoptic models are used as scene models for self-localization of robots, and as object model for training statistical classifiers and for viewpoint selection. In medical image analysis plenoptic models are used in computer-aided endoscopy.

- Motion detection and tracking has been studied in several other projects. Active contours and active rays have been used for real-time object tracking in natural scenes. Iterative state estimation is done using particle filter (e.g. CONDENSATION algorithm).

- Image analysis has been done using a knowledge base represented by a semantic network with parallel or sequential control and an object-oriented implementation.

- Service robots in health care environments and in general autonomous mobile systems is one area of application, where the developed algorithms are implemented and tested in natural scenes. This includes object recognition and tracking and vision-based self-localization.

All implementations of image processing algorithms are now done in an object-oriented programming environment which was designed and realized during the last ten years.


Site generated on Friday, 06 January 2006