Unsupervised textural classification of images using the texture spectrum

Abstract

Pursuing our previous study (Wang and He, Pattern Recognition 23, 905–910 (1990)) where the texture spectrum was proposed for a supervised classification of texture images, an example is presented here of the application of the texture spectrum to an unsupervised textural classification of images. Promising results are obtained when using the algorithm to classify six of Brodatz's natural texture images, demonstrating further the discrimination performance of the texture spectrum.