Low Resolution Image Sampling for Pattern Matching

The paper presents a simulated mobile system that learns to solve the ego- location task in a known environment, in a supervised way, using a very low resolution sampling of the optical array and RBF approximation techniques. The impact of the number of sensors, of their layout, in particular of Sobol sequences with respect to regular grids for a progressively refined sampling of images, and of the complexity of response of each sensing unit has been investigated in an attempt to simplify as much as possible the architecture of the image processing module retaining good localization ability.

[1]  Bernt Schiele,et al.  Recognition without Correspondence using Multidimensional Receptive Field Histograms , 2004, International Journal of Computer Vision.

[2]  George Bosworth Brown.  In Reply: DNA Fingerprinting , 1990 .

[3]  Jan J. Koenderink,et al.  Discrimination thresholds for channel-coded systems , 1992, Biological Cybernetics.

[4]  Roberto Brunelli,et al.  Image Retrieval by Examples , 2000, IEEE Trans. Multim..

[5]  William H. Press,et al.  Numerical recipes in C , 2002 .

[6]  Roberto Brunelli,et al.  Olympus: an ambient intelligence architecture on the verge of reality , 2003, 12th International Conference on Image Analysis and Processing, 2003.Proceedings..

[7]  Edward H. Adelson,et al.  The Design and Use of Steerable Filters , 1991, IEEE Trans. Pattern Anal. Mach. Intell..

[8]  T Poggio,et al.  Regularization Algorithms for Learning That Are Equivalent to Multilayer Networks , 1990, Science.

[9]  T Poggio,et al.  Fast perceptual learning in visual hyperacuity. , 1991, Science.