Abstract

Image feature points are detected as pixels which locally maximize a detectorfunction, two commonly used examples of which are the (Euclidean) imagegradient and the Harris-Stephens corner detector. A major limitation of thesefeature detectors are that they are only Euclidean-invariant. In this work wedemonstrate the application of a 2D affine-invariant image feature pointdetector based on differential invariants as derived through the equivariantmethod of moving frames. The fundamental equi-affine differential invariantsfor 3D image volumes are also computed.

 

Quick Read (beta)