In this paper, texture classification using contourlet transform is proposed. Contourlet transform is a pyramidal directional filter bank, which is a two dimensional extension of wavelet transform. The multiscale and multidirectional property of contourlet transform is very effective in extracting texture features. Zernike moments are applied for its geometrical invariance property as feature selection process for each subband of contorlet transform. Combining contourlet transform and zernike moments are producing good image representative capability because of its properties. Nearest neighbour classifier is used as classifier. For the experimental study brodatz database of textures are used. From the experimental results, it has been observed that Contourlet Transform combined with Zernike moments achieve superior performance than the other well-known models.

Key words: Texture classification, Computer vision, Pattern recognition, Feature Extraction, Contourlet Transform, Zernike moments.