Image Morphing in Deep Feature Spaces: Theory and Applications

DOI10.1007/S10851-020-00974-5arXiv1910.12672MaRDI QIDQ6328015FDOQ6328015

Authors: Alexander Effland, Erich Kobler, Thomas Pock, Marko Rajković, Martin Rumpf 

Publication date: 28 October 2019

Abstract: This paper combines image metamorphosis with deep features. To this end, images are considered as maps into a high-dimensional feature space and a structure-sensitive, anisotropic flow regularization is incorporated in the metamorphosis model proposed by Miller, Trouv'e, Younes and coworkers. For this model a variational time discretization of the Riemannian path energy is presented and the existence of discrete geodesic paths minimizing this energy is demonstrated. Furthermore, convergence of discrete geodesic paths to geodesic paths in the time continuous model is investigated. The spatial discretization is based on a finite difference approximation in image space and a stable spline approximation in deformation space, the fully discrete model is optimized using the iPALM algorithm. Numerical experiments indicate that the incorporation of semantic deep features is superior to intensity-based approaches.

Mathematics Subject Classification ID

Discrete approximations in optimal control (49M25) Geodesics in global differential geometry (53C22) Numerical aspects of computer graphics, image analysis, and computational geometry (65D18) Stability and convergence of numerical methods for ordinary differential equations (65L20) Special approximation methods (nonlinear Galerkin, etc.) for infinite-dimensional dissipative dynamical systems (37L65)

This page was built for publication: Image Morphing in Deep Feature Spaces: Theory and Applications

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6328015)