Neural networks for texture classification

Discusses the application of artificial neural networks to texture analysis. Such networks we used as classifiers in order to distinguish among various classes of texture. A set of experiments was carried out in order to study the network´s classification performance as a function of both its size and the training volume. Another set of experiments was carried out in order to explore the effect of training strategy on the overall performance. The latter shows the effect of the input data arrangement on the learning rate and overall classification performance of the network. Recognition rates of up to 94.4 percent were achieved for tests made on unseen data