Discriminative features for texture description

In this paper, a feature extraction method is developed for texture description. To obtain discriminative patterns, we present a learning framework which is formulated into a three-layered model. It can estimate the optimal pattern subset of interest by simultaneously considering the robustness, discriminative power and representation capability of features. This model is generalized and can be integrated with existing LBP variants such as conventional LBP, rotation invariant patterns, local patterns with anisotropic structure, completed local binary pattern (CLBP) and local ternary pattern (LTP) to derive new image features for texture classification. The derived descriptors are extensively compared with other widely used approaches and evaluated on two publicly available texture databases (Outex and CUReT) for texture classification, two medical image databases (Hela and Pap-smear) for protein cellular classification and disease classification, and a neonatal facial expression database (infant COPE database) for facial expression classification. Experimental results demonstrate that the obtained descriptors lead to state-of-the-art classification performance.

► We present a three-layered learning model for discriminative feature extraction.
► Simultaneously consider robustness, discriminative power and representation capability.
► Generalized model can be integrated with various local binary pattern variants.
► Our approach gives very high classification performances on five datasets.
► Broad applications on texture classification, biomedical diagnosis and expression classification.