Hierarchical neural architecture for visual pattern recognition and reconstruction
Jagath Chandana Rajapakse, Raj S. Acharya · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 1990
A hierarchical self-organizing neural network which can recognize and reconstruct the traces of the previously learned binary patterns is presented. The recognition and reconstruction properties of the network are invariant with respect to distortion, noise, translation, scaling and partial rotation of the original training patterns. If two or more patterns are presented simultaneously, the network pays attention to each pattern selectively. The network can incorporate new training patterns for recognition without loosing its previously learned information. We demonstrate the usefulness of the network in image recognition, reconstruction and segmentation with simulation results.