Fuzzy pattern spectrum as a texture descriptor
M. Ghadiali, Joe C. H. Poon, Wan-Chi Siu · Electronics Letters · 1996
The authors introduce a novel descriptor for texture classification, namely the fuzzy pattern spectrum (FPS). Essentially, the FPS is a novel extension of the traditional morphological pattern spectrum, the significant advantage being its effectiveness in quantifying spatial uncertainty in images. A texture classification experiment is discussed to show the usefulness of the FPS wherein a classification accuracy of 94% is achieved.