Texture Classification Using SVM Fusion

Spectral features of images, such as Gabor filters and wavelet transform can be used for texture image classification. Current methods combining Wavelet and Gabor features blindly, however, it would be better to use each feature-set in independent classifier and fusing based on its score. In this paper, a new method presented which is decision level fusion of SVM classifiers. Experimental results show that decision level fusion usually works better than feature level fusion.

Keywords:

Texture classification , SVM , Wavelet transform , Gabor filter , feature fusion , decision fusion.