ROTATION INVARIANT TEXTURE CLASSIFICATION USING BAMBERGER PYRAMIDS (ThuAmOR5)

Author(s) : Jose Gerardo Rosiles (University of Texas at El Paso, USA) Mark Smith (Purdue University, USA) Russell Mersereau (Georgia Institute of Technology, USA)

Abstract : We study the application of the Bamberger directional filter bank (BDFB) to the problem of rotation invariant texture classification. We explore the use of purely directional decompositions and the use of polar-separable Bamberger pyramids. We obtain comparable classification performance to Gabor-based methods using a smaller feature set.

Menu