Optical Position Converter for Target Tracking and Neural Network

摘要

Optic is widely recognized in information processing for its inherent parallelism. This potential has made optics a promising technology for applications such as pattern recognition~1, neural network~(2,3) and target tracking. The systems developed for these applications often take advantage of the so-called Vander Lugt correlator~4 architecture which permits the recognition of an input pattern with translation invariance. The output is simply a maximum peak correlation where the object to be recognized is located. Cross correlations of lower intensities also appear in the presence of other patterns or of noise.