A novel LBP method for invariant texture classification

Texture classification is a basic task in many applications of machine vision and image processing. Linear Binary Pattern (LBP) methods are among the important categories of invariant texture classification methods. Moreover, Discrete Wavelet Transform (DWT) methods are the other groups of texture classification methods, which attract much attention. LBP features just consider the spatial information of the texture; therefore, this paper proposes a proper combination of the DWT and LBP methods in which we try to improve the ability of LBP methods using multi-resolution analysis. The final results show that the proposed method finely improves the classification rate of the previous and well-known LBP methods for invariant texture classification.