Classification of wooden boards by neural networks and fuzzy rules

Abstract

Fuzzy-neural systems have been applied to many engineering tasks. Fuzzy neurons in pattern classification are extremely useful because they provide a degree of membership information instead of numerical critic values such as "0" (bad) or "1" (good). This paper describes a neural network application for automatic classification of wooden boards. The basic processing unit consists of two types of generic OR and AND neurons structured in a four layer topology.