Optical attentive associative memory with channel dependent nonlinearity in the weight plane

An optical associative processor for two-dimensional image retrieval is described. The system employs variable channel-dependent nonlinearity in the weight plane of the conventional filter-based optical associative memory. Thresholding the filter that is used to store an associative image corresponds to focusing attention on that specific image by altering the strengths of the stored weights in the associative processor. For a severe compression type of nonlinearity, the pixel values of the high spatial frequencies of the filter are increased, which emphasizes the similarity between the input and the selected stored image