Alternative Signatures based on Randomized Neural Network for Texture Classification
Macario Martins Leitao, Jarbas Joaci de Mesquita Sá · 2019
This paper describes two alternative methods for texture signature extraction based on a Randomized Neural Network (RNN), an artificial neural network with a single hidden layer architecture. The proposed signatures are promising ways to increase the discriminating capacity for texture recognition, while also keeping a fast feature building process. Experiments showed that the accuracy of texture classification using the new signatures is higher than other texture description methods in the literature.