Title :
Image representation, texture and the fuzzy n-tuple
Author :
Hepplewhite, L. ; Stonham, T.J.
Author_Institution :
Dept. of Electr. Eng. & Electron., Brunel Univ., Uxbridge, UK
Abstract :
In this paper, the n-tuple pattern recognition method of Bledsoe and Browning is extended to a multiscale representation of the digital image. The new method is based on a fuzzy n-tuple representation of the image´s zero crossing sketch. Filter design and the implications of causality are discussed with respect to the proposed method. Improved parameter selection, classification and noise tolerance are demonstrated by extending n-tuple pattern recognition, to a fuzzy representation
Keywords :
filtering theory; fuzzy set theory; image classification; image recognition; image representation; image texture; noise; causality; classification; digital image; fuzzy n-tuple representation; image representation; image texture; multiscale representation; n-tuple pattern recognition method; noise tolerance; parameter selection; zero crossing sketch; Digital images; Equations; Filters; Frequency; Image edge detection; Image recognition; Image representation; Pattern recognition; Quantization; Stability;
Conference_Titel :
Pattern Recognition, 1998. Proceedings. Fourteenth International Conference on
Conference_Location :
Brisbane, Qld.
Print_ISBN :
0-8186-8512-3
DOI :
10.1109/ICPR.1998.711864
