Evolutionary procedure for the optimisation of a generic texture classifier

Author

Naghdy, Golshah ; Turgut, Alireza

Author_Institution

Dept. of Electr. & Comput. Eng., Wollongong Univ., NSW, Australia

Volume

1

fYear

1997

fDate

28-31 Oct 1997

Firstpage

574

Abstract

A novel evolutionary procedure for the optimisation of a Gabor wavelet-based multi-bank filter network for texture classification is introduced. The performance of the system is evaluated in terms of the classification accuracy. The system starts with a fine mesh of feature extractor filters which result in the highest classification rate. The system is then trimmed down by an evolutionary procedure until a preset performance criterion is reached. The performance of the system is tested using 50 natural textures from the Brodatz album. The results indicates that extremely compact feature vectors with high classification accuracy could be achieved using well-structured test data

Keywords

feature extraction; genetic algorithms; image classification; image texture; performance evaluation; spatial filters; wavelet transforms; Brodatz album; Gabor wavelet-based multi-bank filter network; classification accuracy; classification rate; compact feature vectors; evolutionary procedure; feature extractor filter mesh; generic texture classifier optimization; natural textures; performance evaluation; preset performance criterion; well-structured test data; Computer networks; Computer vision; Continuous wavelet transforms; Feature extraction; Filter bank; Frequency; Gabor filters; Humans; Spatial resolution; Visual system;

fLanguage

English

Publisher

ieee

Conference_Titel

Intelligent Processing Systems, 1997. ICIPS '97. 1997 IEEE International Conference on

Conference_Location

Beijing

Print_ISBN

0-7803-4253-4

Type

conf

DOI

10.1109/ICIPS.1997.672849

Filename

672849