Title :
Texture classification using kernel independent component analysis
Author :
Jian Cheng ; Qingshan Liu ; Hanqing Lu
Author_Institution :
Inst. of Autom., Chinese Acad. of Sci., Beijing, China
Abstract :
We propose a novel method, kernel independent component analysis (KICA), for texture features extraction. The texture images are first mapped into a higher-dimensional implicit feature space. Then a set of nonlinear basis functions are learned using KICA. The feature vectors are obtained by projected the texture images onto the basis functions. Comparison experiments between KICA and the other two classic methods: Gabor filters and ICA, are performed. The results indicate that the KICA is an efficient approach for texture classification.
Keywords :
feature extraction; image classification; image texture; learning (artificial intelligence); nonlinear functions; vectors; Gabor filters; feature vectors; higher dimensional implicit feature space; kernel ICA; kernel independent component analysis; learning; nonlinear basis functions; texture classification; texture features extraction; texture images; Automation; Decorrelation; Educational institutions; Feature extraction; Gabor filters; Independent component analysis; Kernel; Laboratories; Pattern recognition; Principal component analysis;
Conference_Titel :
Pattern Recognition, 2004. ICPR 2004. Proceedings of the 17th International Conference on
Print_ISBN :
0-7695-2128-2
DOI :
10.1109/ICPR.2004.1334231