Affine Invariants Based on Covariance Matrix
Yishu Liu · Journal of Chinese Computer Systems · 2007
The feature-based recognition of affine-deformed objects is important and challenging in pattern recognition and computer vision,finding efficient features which are invariant under affine transformation is the key to solving this problem.This paper concentrates on deriving such features as mentioned above,i.e.affine invariants.The covariance matrix of a given image is computed first,then its' eigenvalues and eigenvectors are calculated, with that a set of ellipses with the same center are constructed,lastly a set of affine invariants are derived based on image compactification theory and some properties of affine transformation.Numerical experiments are carried out and show that the performance of our affine invariants is promising.