A combined neural network approach for texture classification

-In this article, we present a two-stage neural network structure that combines the characteristics o f selforganizing map ( S O M ) and multilayer perceptron (MLP) for the problem of texture classification. The texture features are extracted using a multichannel approach. The channels comprise o f a set o f Gabor filters having different sizes, orientations, and frequencies to constitute N-dimensional feature vectors. S O M acts as a clustering mechanism to map these N-dimensional feature vectors onto its M-dimensional output space, where in our experiments, the value o f M was taken as two. This, in turn, forms the feature space from which the features are f ed into an M L P for training and subsequent classification. It is shown that the disadvantage of using Gabor filters in texture analysis, namely, the higher dimensionality o f the Gaborian feature space, is overcome by the reduction in the dimensionality of the feature space achieved by SOM. This results in a significant reduction in the learning time of M L P and hence the overall classification time. It is found that this mechanism increases the interclass distance (average distance among the vectors o f different classes) and at the same time decreases the intraclass distance (average distance among the vectors o f the same class) in the feature space, thereby reducing the complexity o f classification. Experiments were performed on images containing tiles o f natural textures as well as image data from remote