Decomposition and Recognition of Non-orthogonal Interacting Features Using an Sofm Neural Network

Recognition of interacting features has been a difficult task in many existing feature-recognition systems. The unique topological patterns of isolated features change drastically when they interact. Hence many surfacebased methods encounter problems in accommodating these changes in their generic feature definitions. Recently, much effort has been concentrated on the volumetric approach. However, many of these systems suffer from a problem of combinatorial explosion as the interaction between features becomes more complex. This paper presents a simple and robust system, in which the non-orthogonal interacting features are decomposed into non-orthogonal regions using a Kohonen self-organizing feature map (SOFM) neural network. The feature patterns in these non-orthogonal regions are then used as input in a multilayer feedforward neural network to recognize the features. Self-organization, competitive learning and the clustering of data are some of the SOFM’s attributes, exploited in this work to deal with interacting features.