p.246
p.252
p.257
p.264
p.271
p.276
p.281
p.285
p.290
Human Ear Recognition Using Hybrid Filter and Supervised Locality Preserving Projection
Abstract:
To solve the difficult problem of human ear recognition caused by variety of ear angle, a novel method which combines hybrid filter with supervised locality preserving projection (SLPP) is proposed. The ear image is firstly filtered by Log-Gabor filter which is constructed with 5 scales and 8 orientations. The important parameters of Log-Gabor filter are selected through experiments. To form effective and discriminative feature too many Log-Gabor coefficients are reduced by discrete cosine transform. Lastly feature is constructed by SLPP to discovery geometrical rules. Experimental results show that compared with the traditional methods, the proposed method obtains higher recognition rate, and is robust to multi-pose of ear recognition.
Info:
Periodical:
Pages:
271-275
Citation:
Online since:
June 2012
Authors:
Price:
Сopyright:
© 2012 Trans Tech Publications Ltd. All Rights Reserved
Share:
Citation: