Complex wavelet based texture classification

Texture classification plays an important role in image analysis and understanding. The real wavelet based methods are deprived of the significant benefits of the phase information. Thus, the complex wavelet should be taken into account. This paper will combine the phase information and the magnitude information of the complex wavelet coefficient into a real measure to describe the intensity variation of a texture, and then model the measure with the real generalized Gaussian distribution (GGD). The model parameters serve as the texture feature during the classification. The experimental results on two benchmark texture databases demonstrate the superior performance of the new texture feature.