Rotation Invariant Texture Classification Using Bamberger Pyramids
Jose Gerardo Rosiles, Mark J. T. Smith, Russell M. Mersereau · 2005
We study the application of the Bamberger directional filter bank 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.