Invariant recognition by feedback network

Author

Zhang, Yongjun ; Chen, Zongshi

Author_Institution

Inst. of Electron., Acad. Sinica, Beijing, China

Volume

3

fYear

1993

fDate

25-29 Oct. 1993

Firstpage

2185

Abstract

A method of pattern recognition using a multilayered feedback neural network is described. Experiments were carried out for ISAB radar image in a frame recognition using neural network. The problem of pattern recognition using a feedback neural network is as follows. First, the input image is Fourier transformed. The absolute of translation of the input image is invariant against translations in the plane. In the second step, the previously transformed image is transformed again with complex-log mapping

Keywords

Fourier transforms; content-addressable storage; correlation methods; image recognition; multilayer perceptrons; recurrent neural nets; Fourier transformation; ISAB radar image; complex-log mapping; correlation associative memory network; frame recognition; invariant recognition; iteration; multilayered feedback neural network; pattern recognition; rotation-and-scale invariant image; Art; Feature extraction; Feedback; Fourier transforms; Image recognition; Multi-layer neural network; Neural networks; Neurofeedback; Pattern recognition; Radar imaging;

fLanguage

English

Publisher

ieee

Conference_Titel

Neural Networks, 1993. IJCNN '93-Nagoya. Proceedings of 1993 International Joint Conference on

Print_ISBN

0-7803-1421-2

Type

conf

DOI

10.1109/IJCNN.1993.714159

Filename

714159