Effect of wavelet bases in texture classification using a tree-structured wavelet transform

A number of algorithms have been developed in the past for texture analysis and classification using the wavelet transform. In this paper we explore the effect of different wavelet bases on the classification performance. We use the algorithm developed by T. Chang and C.-C. Jay Kuo (see IEEE Trans. Image Process., vol.2, no.4, p.429-41, Oct. 1993) for texture analysis and classification with a tree-structured wavelet transform. The performance of the wavelet bases is measured in terms of the sensitivity and the selectivity for the classification of natural textures. The classification results show that the choice of wavelet basis has a considerable effect on the classification performance.