Title :
Generalized pattern spectra sensitive to spatial information
Author :
Wilkinson, Michael H E
Author_Institution :
Inst. for Math. & Comput. Sci., Groningen Univ., Netherlands
Abstract :
Morphological pattern spectra computed from granulometries are frequently used to classify the size classes of details in textures and images. An extension of this technique, which retains information on the spatial distribution of the details in each size class is developed. Algorithms for computation of these spatial pattern spectra for a large number of granulometries on binary images are presented.
Keywords :
image texture; mathematical morphology; pattern classification; spectral analysis; binary images; granulometries; image textures; morphological pattern spectra; pattern classification; spatial information; Image reconstruction; Image segmentation; Image texture analysis; Information analysis; Mathematics; Microscopy; Pattern analysis; Pixel; Shape; Soil;
Conference_Titel :
Pattern Recognition, 2002. Proceedings. 16th International Conference on
Print_ISBN :
0-7695-1695-X
DOI :
10.1109/ICPR.2002.1044579
