Research On Fuzzy Morphological Bidirectional Associative Memories

Associative memory is one of important capabilities of human brain cell. In recent years, morphological bidirectional associative memories(MBAM) presented by GX.Ritter is a latest advanced in this aspect. It overcomes disadvantage of the traditional associative memory model which have less storage capabilities, need many iterations and can not even converge. MBAM circumvents these disadvantages and obtains the success binary image processing.Based on MBAM, fuzzy morphological bidirectional associative memories(FMBAM) is presented in this paper and the conditions that guarantee perfect bidirectional associative memories are derived. Its perfect bidirectional associative memory capability comes from the taken multiplication/max.mun operations. What is more important due to the fact that FMBAM can storage fuzzy rules and extend MBAM's storage capacity. Therefore it provides a new way for associative memory application. In order to effectively suppress random noise, dynamic-kernel based FMBAM is presented. This paper analyzes characteristics of dynamic kernel vectors and indicates its validity. Meanwhile this paper gives the method and steps of searching optimal kernel vectors. Thus FMBAM can be used to suppress random noise in binary image. Accordingly, it is successfully used in pattern recognition issues of binary image processing. Our experimental results indicate that it obviously outperforms other traditional associative memory models.FMBAM's applications in gray/color images are also investigated. Our initial results indicate its feasibility. How choose kernel vectors is the main difficulty in gray images precessing. In order to solve the problem, a novel search mothed for dynamic kernelis studies. When FMBAM is used to color images, computational time is a heavy burden. To sove the computational time, we first choose three kernels respectively for three different colors, and then integrate them together. Our experiment resulrs demonstrate the...