Empirical Mode Decomposition for Rotation Invariant Texture Classification

A novel and effective scheme for rotation invariant texture classification is presented using an adaptive and approximately orthogonal filtering process Bidimensional Empirical Mode Decomposition (BEMD). The extraction of rotation invariant feature for a given image involves BEMD and circular zones. A feature vector extracted from circular zones of intrinsic mode function (IMF) is constructed for rotation invariant texture classification. In the experiments, we use rotation invariant feature to classify a set of 25 distinct natural textures selected from the Brodatz album. The experimental results show that the effectiveness of the proposed classification scheme compared with other classification methods.