Classification of rice grains using fuzzy artmap neural network

Author

Wee, Chong- Yaw ; Paramesran, Raveendran ; Takeda, F. ; Tsuzuki, Toshihiro ; Kadota, Hiroshi ; Shimanouchi, S.

Author_Institution

Dept. of Electr. & Telecommun. Eng., Malaya Univ., Kuala Lumpur, Malaysia

Volume

2

fYear

2002

fDate

2002

Firstpage

223

Abstract

In this paper, a scaled invariant Zernike moment based feature extractor has been used to extract the relevant information from rice grain images for the purpose of classification. An incremental supervised learning and multidimensional map neural network, called fuzzy artmap (FA), has been proposed to reduce the learning time while maintaining high accuracy. A fast computation technique that uses the higher order Zernike polynomials to derive the lower order Zernike polynomials has been proposed to improve the computation speed of Zernike moments in real time applications.

Keywords

ART neural nets; Zernike polynomials; feature extraction; food processing industry; fuzzy neural nets; image classification; method of moments; scaling phenomena; FA accuracy; fast computation techniques; fuzzy artmap neural networks; higher/lower Zernike polynomials; incremental supervised learning; learning time reduction; multi-dimensional map neural networks; real time applications; rice grain classification; rice grain image information extraction; rotational invariant features/properties; scaled invariant Zernike moment-based feature extractors; Backpropagation algorithms; Communication industry; Data mining; Feature extraction; Fuzzy neural networks; Multilayer perceptrons; Neural networks; Polynomials; Supervised learning; Systems engineering and theory;

fLanguage

English

Publisher

ieee

Conference_Titel

Circuits and Systems, 2002. APCCAS '02. 2002 Asia-Pacific Conference on

Print_ISBN

0-7803-7690-0

Type

conf

DOI

10.1109/APCCAS.2002.1115197

Filename

1115197