Multisensor information fusion in neural networks on the basis of diffraction optics

The problem of sensor data fusion is considered from the viewpoint of a cognitive approach and recent investigations into the fusion of information of different modality in biological systems. The physioassociative temporal sensor integration algorithm is used. This algorithm accounts for a variety of emergent processing loops for prediction, feature extraction, matching, and search in static and dynamic databases. It is shown how processing loops of different type can be realized in hybrid optoelectronic neural networks with the use of diffraction optics and holography.