Hello all,

I'm interested in 2D discrete transforms (such as Discrete Wavelet Transforms, [Curvelets][1], Ridgelets, Beamlets etc.) that are invariant to rotations **only**, and that effectively compress the input 2D signal efficiently (in terms of computational complexity).

In other words, I'm interested in 2D discrete transforms that output the same transformation for arbitrary 2D rotations of the input, but that are not invariant to changes in scale, translation, or any other rotation-independent attribute of the input. Ideally, these transforms should compress the input 2D signal as much as possible in terms of information entropy, i.e. necessary bits to represent the output.

As additional context to my question, I am planning on using such transforms in the domain of image processing/computer vision to train a classifier on instances of objects that might appear rotated arbitrarily around the image center point.

Thank you

----------


  [1]: http://www.curvelet.org/