Morphological neural networks and image algebra in artificial perception systems
Carmen Paz Suárez Araujo, Gerhard X. Ritter · 1992
generalized matrix product obtaining new computational models as morphological neural networks (MNN).In this paper we propose a theoretic approach on the invariant perception. We also show that image algebracan be used not only in the field ofimage processing but in other areas related to artificial perception systems.Our study is based on both a general theory of neural network and the invariant perception by the cortextheory. The neural structures that we propose uphold both the architecture and functionality ofthe cortex.We present a neural network model for computing auditory homothetic invariances in accordance with ageneral framework in image algebra. The neuronal synthesis of this model is obtained using MNN theory withthe binary operations the maximum and the multiplication in the neural network formulation. We also proposea second model which is achieved introducing a simple logarithmic transformation in the current model.In addition we propose an alternative MNN for computing homothetic invariances which arise from how theproblems are formulated in the systems of artificial vision. This last neural network is appropriate to computevisual invariances when we process patterns defined in two dimension spaces.1. INTRODUCTIONThe fascination of neural network computing arises in part from its relationship to the brain. Claims havebeen made in recent years that artificial neural networks (ANNs) explain the basic mechanism and dynamics ofthe brain, but vast differences exist between biological neural network (BNN), the cerebral cortex for example,and artificial neural network, both in architecture and capabilities'.