Noise-resistant affine skeletons of planar curves

A new definition of affine invariant skeletons for shape representation is introduced. A point belongs to the affine skeleton if and only if it is equidistant from at least two points of the curve, with the distance being a minima and given by the areas between the curve and its corresponding chords. The skeleton is robust, eliminating the need for curve denoising. Previous approaches have used either the Euclidean or affine distances, thereby resulting in a much less robust computation. We propose a simple method to compute the skeleton and give examples with real images, and show that the proposed definition works also for noisy data. We also demonstrate how to use this method to detect affine skew symmetry.