Spline-Based Image Registration
International Journal of Computer Vision
| Hi-index | 0.00 |
Deformations due to relative motion between an observer and an object may be used to infer 3-D structure. Up to first order these deformations can be written in terms of an affine transform. This paper considers the problem of measuring the affine transform locally between two image patches using weighted moments of brightness. It is shown that the moments of image patches are related through functions of affine transforms. Finding the weighted moments is equivalent (for the purposes of measuring the affine transform) to filtering the images with gaussians and derivatives of gaussians. In the special case where the affine transform can be written as a scale change and an in-plane rotation, the zeroth and first moment equations are solved for the scale. Experimentally, the method seems to be robust.