Institute of Integrated Sensor Systems

Prof. Dr.-Ing. Andreas König







Student Theses




QuickCog - Cognitive Systems Design Environment

Andreas König, Michael Eberhardt, and Robert Wenzel

QuickCog was devised as PC based (Windows 95/98/NT/2k/XP) rapid and efficient system design tool for automated visual inspection tasks, e.g. in medical product inspection or electronics manufacturing, and general pattern recognition applications. The key research idea was to conceive a modular environment, that can be adapted to new or modified task with little or no expert knowledge at all employing transparent user interaction and optimization. Fast and transparent, and, thus, accelerated or automated design of cognitive systems is provided by the following key features:

Visual programming
Interactive menu-driven creation of arbitrary sample sets, ROI specification (object partitioning/object scanning), and complete image or partitioned object consistent preclassification for whole sample set
Easy access to sample sets and processing methods
Powerful and proven methods of image processing, feature computation, classification and evaluation (classification rates, confusion matrices, and error visualisation) in a single system
Feature space visualisation and interactive analysis by dimension reducing mappings and interactive feature maps, assessment functions for system discriminance, method selection, and parameter settings
Automatic feature selection by parametric and nonparametric methods
Efficient and easy to use classification methods, including centroid classifier, nearest-neighbor-techniques, hypersphere classifier, rule-based and hierarchical classifiers as well as various neural networks.
Instant deployment facility of the designed cognitive system on industrial PC/104+ standard as well as any PC system (486, 8 MB, 300 MByte Disk minimum requirements). SmartEYE deployment is considered as next step of integration.

QuickCog Adaptive System Architecture:

Future extensions will enhance QuickCog by the following features:

Wavelet- and Gabor-filters
Colour image processing
Image sequence analysis
Enlarged Neural-Network-Toolbox
Development assistant for user specific system enhancements in visual-C++ (C/C++-Plug-In)
as well as by improvements of the graphical user interface to facilitate application for a broad community of industrial or educational users.

QuickCog was introduced at CeBIT'98 in Hannover (March 18-25, 1998) and is commercially available since fall 1998. An English version is scheduled for end 1999. Further presentations of QuickCog in 1998/99/00 were at:

EUFIT'98, Aachen, September 6-9
VISION'98, Stuttgart, Sept. 28th - Okt. 1rst
ELECTRONICA'98, München, Oct.
COMTEC'98, Dresden, Nov. 25-28
Hannover Fair 1999, April 18-24, ''Forschungsland Sachsen'', Halle 18, 1.OG, J10
Sensor 99, May 18-20, ''Forschungsland Sachsen'', Nürnberg
VISION 99, October, Phytec Meßtechnik GmbH, Stuttgart, QuickCog 2.0 Release
Embedded Systems 2000, 16.- 18. Februar 2000, Nürnberg, Stand Phytec Meßtechnik GmbH
VISION 2000, 18.- 20. Oktober 2000, Stuttgart, Stand Phytec Meßtechnik GmbH, QuickCog 2.5 Release
electronica 2000, 21.- 24. November 2000, München, Stand Phytec Meßtechnik GmbH
SPS/PC/DRIVES 2000, 28. - 30. November 2000, Nürnberg, Stand Phytec Meßtechnik GmbH
Hannover Fair 2001, April 22-28, ''Forschungsland Sachsen'', Halle 18, 1.OG, M16

Further Product Information:

  Status:   concluded, duration 10/1997-09/2002
Design automation of intelligent systems is pursued in several follow-up projects dedicated to sensor signal & image processing.
  Partner:   PHYTEC Messtechnik GmbH, Robert-Koch-Strasse 39, 55129 Mainz
  Financing:   PHYTEC Messtechnik GmbH, Robert-Koch-Strasse 39, 55129 Mainz; self-funded
  Contact:   Prof. Dr.-Ing. Andreas König
  Key Contributors:   Michael Eberhardt, Robert Wenzel, Andreas König and numerous other students
      König, A., Eberhardt, M., Wenzel, R.: A Transparent and Flexible Development Environment for Rapid Design of Cognitive Systems . In Proceedings of the EUROMICRO'98 Conference, Workshop Computational Intelligence, Väasteras, Sweden, August 25-27, 1998
      König, A.: A Survey of Methods for Multivariate Data Projection, Visualisation and Interactive Analysis. In Proceedings of the 5th International Conference on Soft Computing and Information/Intelligent Systems IIZUKA'98, pp. 55-59, Iizuka, Fukuoka, Japan, October 16-20, 1998.
      Eberhardt, M., Wenzel, R., König, A..: Efficient Application of Neural Networks in a Self-Learning System for Visual Inspection. Proc. of the Fourth Int. Workshop on Neural Networks in Applications NN'99, Magdeburg, Germany, pp. , March, 1999.
      König, A., Herenz, A., Wolter, K.: Application of Neural Networks for Automated X-Ray Image Inspection in Electronics Manufacturing. In Proc. of the International Work-Conference on Biological and Artificial Neural Networks IWANN'99, Alicante, Spain, pp. 588-595, June, 1999.
      König, A., Eberhardt, M., Wenzel, R.: QuickCog Self-Learning Recognition System - Exploiting machine learning techniques for transparent and fast industrial recognition system design. In Image Processing Europe, PennWell, pp. 10-19, Sept./Oct. Issue, 1999.
      König, A.: Dimensionality Reduction and Interactive Visualization of Multivariate Data - Methods, Tools, Applications. Tutorial Slides presented at 6th on-line World Conference on Soft-Computing WSC6 (pdf-file of 107 slides, 2.5 MB, pdf-file of accompanying paper, 5.3 MB), Fraunhofer IPK, , September, 2001.