A Fast Method to Decide Kernel Patterns for Morphological Associative Memory

Two-step Morphological Associative Memories (MAMs) using kernel patterns have a lot of desirable features. It is, however, very difficult to decide kernel patterns when the number of training patterns to be stored is large and when training patterns have no unique feature bits. In this paper, we propose a novel method of constructing kernel patterns for the two-step MAMs. We derive the proposed method by examining the relation between characteristics of kernel patterns and outputs of MAMs. The proposed method has the following features: (1)It can construct morphological memory from original patterns; (2)It can converge very fast because it does not require trial and error to decide kernel patterns; (3)It keeps the inherent features of MAMs such as unlimited storage capacity, robustness for both erosive and dilative noise, and so on. Computer simulation results show that the proposed method is more than 10 million times faster than the conventional method using trail and error.