Pattern recognition using dual-tree complex wavelet features and SVM

Author

Chen, G.Y. ; Xie, W.F.

Author_Institution

Dept. of Comput. Sci., Concordia Univ., Montreal, Que.

fYear

2005

fDate

1-4 May 2005

Firstpage

2053

Lastpage

2056

Abstract

A novel descriptor for pattern recognition is proposed by using dual tree complex wavelet features and SVM. The approximate shift-invariant property of the dual tree complex wavelet and its good directional selectivity in 2D make it a very appealing choice for pattern recognition. Recently, SVM has been shown to be very successful in pattern recognition. By combining these two tools we find that better recognition results are obtained. The recognition rate of similar patterns improves from 94% to 99.25% for both the Gaussian kernel and the wavelet kernel. It is concluded that the dual-tree complex wavelet should be used instead of the scalar wavelet for pattern recognition applications

Keywords

Gaussian processes; pattern recognition; support vector machines; trees (mathematics); wavelet transforms; Gaussian kernel; SVM; dual-tree complex wavelet features; pattern recognition; scalar wavelet; shift-invariant property; Computer science; Feature extraction; Fourier transforms; Handwriting recognition; Kernel; Multi-layer neural network; Neural networks; Pattern recognition; Support vector machines; Wavelet transforms;

fLanguage

English

Publisher

ieee

Conference_Titel

Electrical and Computer Engineering, 2005. Canadian Conference on

Conference_Location

Saskatoon, Sask.

ISSN

0840-7789

Print_ISBN

0-7803-8885-2

Type

conf

DOI

10.1109/CCECE.2005.1557390

Filename

1557390