Pde Based Features for Texture Analysis Using Wavelet Transform

In the present paper, a novel method of partial differential equation (PDE) based features for texture analysis using wavelet transform is proposed. The aim of the proposed method is to investigate texture descriptors that perform better with low computational cost. Wavelet transform is applied to obtain directional information from the image. Anisotropic diffusion is used to find texture approximation from directional information. Further, texture approximation is used to compute various statistical features. LDA is employed to enhance the class separability. The k-NN classifier with tenfold experimentation is used for classification. The proposed method is evaluated on Brodatz dataset. The experimental results demonstrate the effectiveness of the method as compared to the other methods in the literature.