Anwendung Neuronaler Netze für die Texturklassifikation in gemusterten Materialien

The following article concerns with some considerations and results in using neural networks for classification of textile materials. Advantages and disadvantages of using a backpropagation-classifier will be described by the help of an easy modelling of its functionality. In a practical example - surface inspection in textile materials - feature sets calculated from the co-occurrence-matrix are to be classified by a neural network. This proofs the possibility of using neural network in detection of distortions in industrial surfaces.