Abstract

Our research group has recently begun an investigation on the relationship between the field of digital image processing and the field of artificial neural network. So far, we found much commonalities between the two fields in the algorithms they develop, although they frequently use different terminologies. For example, we found that image restoration is closely related to auto-associative memory, while pattern recognition is closely related to hetero-associative memory. As another example, background suppression in image processing is quite similar to attentive associative memory, while match-filtering process can be performed by accretive associative memory. In this paper we concentrate on the comparison of adaptive pattern recognition and iterative image restoration algorithms with associative mappings.

© 1987 Optical Society of America