Texture extraction capabilities of Multi-layer Perceptron

Davi P. SantosJoão Batista Neto

This work explores the Multi-layer Perceptron's inference capabilities to detect textured relationships of pixels belonging to a squared neighbourhood. Although hidden in the neuron connections, these relationships lend the neural network the necessary discriminant power to classifypatterns. Results similar to those involving the combination co-occurrence matrices-MLP havebeen obtained for supervised image segmentation. The work also sheds the basis for an unsupervisedtextured segmentation approach involving an autoassociative neural network and a Self-Organizing Map.

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato:
     Mantida por: