A subspace approach for matching 2D shapes under affine distortions

This paper presents a subspace approach to matching a pair of 2D shapes, and estimating the affine transformation that aligns the two 2D shapes. In the proposed method, by considering each shape as a 2D signal, one shape is projected onto the subspace spanned by the other, and the affine transformation is estimated by minimizing the projection error in the subspace. The proposed method is fast, easy to implement, and with a clear physical interpretation. Furthermore, it is robust to noise due to the merit of the subspace method. The proposed approach has been tested for registration accuracy, computation time, and robustness to noise. Its performance on synthetic and real images is compared with the state-of-the-art reference algorithms. The experimental results show that our approach compares favorably to the reference methods, in terms of registration accuracy, computation speed, and robustness.