Prerational Intelligence: Adaptive Behavior and Intelligent Systems Without Symbols and Logic, Volume 1, Volume 2 Prerational Intelligence: Interdisciplinary Perspectives on the Behavior of Natural and Artificial Systems, Volume 3 pp 123-150 | Cite as
Invariances in Visual Systems of Robots and Animals: Functional and Architectural Consequences
Speaking simultaneously about robots and animals in the title of this contribution seems rather provocative. We are still far away from understanding biological brains and we feel it is useless to attempt efforts to imitate them. On the other hand, the performance of biological systems shows that there exist solutions for all the outstanding features in the context of invariances. Our human visual system impresses by the almost unlimited number of recognition problems which we can easily solve without being conscious of the difficulty. We can rapidly learn objects without training sequences, and we are able to recognize them robustly “at a glance”. We are not even conscious that there could be problems with invariances or 3D vision. Without any zoom we are able to get a panoramic view of an extended scene, and somewhat later we can concentrate on a small detail in highest resolution. Without any doubt, all these features belong to the field of prerational intelligence.
KeywordsFeature Vector Receptive Field Complex Cell Spatial Frequency Channel Contour Element
Unable to display preview. Download preview PDF.
- Büker, U. , & G. Hartmann (1996). Knowledge based view control of a neural 3-D object recognition system. Proceedings of 13th Conference on Pattern Recognition, Vol. IV (pp. 24–29). IEEE Computer Society Press.Google Scholar
- Büker, U. (1998). Hybrid object models: Combining symbolic and subsymbolic object recognition strategies. In N. Callaos, O. Omolayole, & L. Wang (eds. ), Proceedings of the 4th Int. Conference on Information Systems, Analysis and Synthesis, Vol. 1 (pp. 444–451). Orlando, FL: International Institute of Informatics and Systemics.Google Scholar
- Dunker, J. , G. Hartmann, & M. Stöhr (1996). Single view recognition and pose estimation of 3D objects using sets of prototypical views and spatially tolerant contour representations. Proceedings of 13th Conference on Pattern Recognition, Vol. IV (p. 14–18). IEEE Computer Society Press.CrossRefGoogle Scholar
- Hartmann, G. (1983). Processing of continuous lines and edges by the visual system. Biological Cybernetics 47, 43–50.Google Scholar
- Hartmann, G. (1991a). Hierarchical neural representation by synchronized activity: A concept for visual pattern recognition. In J. G. Taylor, E. R. Caianiello, R. M. J. Cotterill, & J. W. Clark (eds. ), Neural network dynamics (pp. 356–370). London: Springer.Google Scholar
- Hartmann, G. (1991b). Learning in a closed loop antagonistic network. In T. Kohonen, K. Mäkisara, O. Simula, & J. Kangas (eds. ), Proceedings of the ICANN-91 (pp. 239–244). Amsterdam: Elsevier/North-Holland.Google Scholar
- Hartmann, G. , S. Drüe, K. O. Kräuter, & E. Seidenberg (1993). Simulations with an artificial retina. Proceedings of the World Congress on Neural Networks1993, Vol III (pp. 689–694). Hillsdale, NJ: Lawrence Erlbaum Ass.Google Scholar
- Hartmann, G. , & S. Drüe (1994). Why synchronization? An attempt to show quantitative advantage. Proceedings World Congress on Neural Networks1994, Vol. I (pp. 581–586). Hillsdale, NJ: Lawrence Erlbaum Ass.Google Scholar
- Hubel, D. H. , & T. N. Wiesel (1959). Receptive fields of single neurons in the cat’s striate cortex. Journal of Physiology 148, 574–579.Google Scholar