Regularization of orthonormal vector sets using coupled PDE's

Abstract

We address the problem of restoring, while presenting possible discontinuities, fields of noisy orthonormal vector sets, taking the orthonormal constraints explicity into account. We develop a variational solution for the general case where each image feature may correspond to multiple n-D orthogonal vectors of unit norms. We first formulate the problem in a new variational framework, where discontinuities and orthonormal constraints are preserved by means of constrained minimization and /spl Phi/-function regularization, leading to a set of coupled anisotropic diffusion PDE. A geometric interpretation of the resulting equations, coming from the field of solid mechanics, is proposed for the 3D case. Two interesting restrictions of our framework are also tackled: the regularization of 30 rotation matrices and the direction diffusion (the parallel with previous works is made). Finally, we present a number of denoising results and applications.