Geometric PCA of Images
Jérémie Bigot, Raúl Gouet, Antonio Manuel López · SIAM Journal on Imaging Sciences · 2013
We describe a method for analyzing the principal modes of geometric variability of images. For this purpose, we propose a general framework based on the use of deformation operators for modeling the geometric variability of images around a reference mean pattern. In this setting, we describe a simple algorithm for estimating the geometric variability of a set of images. Some numerical experiments on real data are proposed for highlighting the benefits of this approach. The consistency of this procedure is also analyzed in statistical deformable models.