Neural abstraction pyramid: a hierarchical image understanding architecture
Sven Behnke, Raúl Rojas · 2002
A hierarchical neural architecture for image interpretation is proposed, which is based on image pyramids and cellular neural networks inspired by the principles of information processing found in the visual cortex. The algorithms for this architecture are defined in terms of local interactions of processing elements and utilize horizontal as well as vertical feedback loops. The goal is to transform a given image into a sequence of representations with increasing level of abstraction and decreasing level of detail. A first application, the binarization of handwriting, has been implemented and shown to improve the acceptance rate of an automatic ZIP-code recognition system without decreasing its reliability.