CSDD Features: Center-Surround Distribution Distance for Feature Extraction and Matching

Motivation A new interest region operator and feature descriptor called Center-Surround Distribution Distance (CSDD) is based on comparing feature distributions between a central foreground region and a surrounding ring of background pixels. In addition to finding light(dark) blobs surrounded by a dark(light) background, CSDD also detects blobs with arbitrary color distribution that “stand out” perceptually because they look different from the background. CSDD detection repeatability is evaluated and compared with other state-of-the-art approaches using a standard dataset, while use of CSDD features for image registration is demonstrated using a RANSAC procedure for affine image matching. We compared repeatability scores between the circular cCSDD detector, an elliptical eCSDD detector, and five other state-ofthe-art detectors (Harrisand Hessian-affine, MSER, edge-based (EBR), and intensity extrema-based (IBR)) for the eight image sequences from the standard affine covariant region detector evaluation dataset, available at http://www.robots.ox.ac.uk/~vgg/research/affine/