Bioinformatics

Issue

Bioinformatics, pattern recognition and image processing, proteomics

Personal Data

Hans Burkhardt, Prof. Dr.-Ing.

Department

Pattern Recognition and Image Processing, Department of Computer Science, University of Freiburg

Know-How

Fields of research:

• Invariant theory for 2D images and 3D objects.


Methods of invariant pattern recognition and visual examination of gray value and binary images.



• Design of Classifiers.


Development and enhancement of efficient classifiers. Our primary research focus concentrates on Support Vector Machines (SVM).



• Fundamental problems in Computer Vision.


Besides projective geometry we also do research on camera models and multi-image geometrical problems.



• Motion and depth estimation based on stereo and multiple camera views as well as image sequences.


Motion and depth estimation for dealing with stereo- and multi camera view records of image sequences in computer vision.



• Digital image filtering and restoration algorithms.


Non-linear and multi sensor algorithms in the field of image filtering and restoration.



• Multi-spectral image analysis of processes in chemical engineering.


Monitoring and control of combustion processes and flow structures of circulating fluidized beds.



• Development of fast and parallel algorithms.


Development of fast and parallel algorithms and multi-process structures in image processing and pattern recognition.

Experience

See overview "fields of research and projects"

Publications

See publication list

Projects

See project list

Technology Offer

Protein Shape Retrieval - Fast Search in Protein Fold Databases

The department of Computer Sciences has built a software interface for retrieving structurally similar proteins in large protein fold databases. Similar proteins can be retrieved in a few seconds out of a database with 30,000 protein structures. We designed a new algebraic method for structural comparison between proteins based on invariant features computed by group integration with spherical harmonics. Our goal is to achieve accurate and fast lassification without alignment by using intrinsic, pose invariant features of the fold structure. In comparison with existing software (e.g. DALI), our method is 10,000 times faster only losing a few percent in precision.

Please find complete Technology Offer and Scientific Informations here