Texture Classification Using Shift-Invariant Wavelet Packet Decomposition

This paper proposes a high performance texture classification method using dominant energy features based on shift-invariant wavelet packet coefficients obtained by 2D shift-invariant wavelet packet decomposition. Experiments employing a reduced feature set show that the proposed method involves a relatively small classification time while still achieving a high accuracy rate (95.6%) for classifying twenty classes of natural texture images.