Description of Rotation-Invariant Textures using Local Binary Pattern Features

Abstract

classification is one of the most interesting research topics in the field of computer vision. This paper aims at classifying static as well as dynamic textures (DT). Uniform Local binary pattern (LBP) is a combination of structural and statistical analysis model for classification of both static and dynamic textures. The LBP Histogram Fourier (LBP-HF) uses Fast Fourier Transform (FFT) for calculating global rotation invariant of LBP histogram for static texture classification. For dynamic texture classification, LBP-TOP method is used, which computes LBP from three orthogonal planes. It combines both motion and appearance together for classification. The results show that, LBP-HF outperforms uniform LBP as well as basic rotation invariant LBP method for static-textures and LBP-TOP gives higher accuracy for dynamic-textures.