We present a fast and high order method for the problem of Image Registration, using Optimal Transport and the Isogeometric Analysis paradigm. Our method is based on the resolution of the Monge-Ampère equation and ensures the one-to-one property. In addition, the use of B-Splines allows to create a map that can be evaluated everywhere, and reduces the number of degrees of freedom needed to store the constructed (gradient) map, by using e.g. high order B-Splines functions.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Isogeometric Analysis For Image Registration and Segmentation Using Optimal Transport Problem

  • Mustapha Bahari,
  • Abderrahmane Habbal,
  • Ahmed Ratnani

摘要

We present a fast and high order method for the problem of Image Registration, using Optimal Transport and the Isogeometric Analysis paradigm. Our method is based on the resolution of the Monge-Ampère equation and ensures the one-to-one property. In addition, the use of B-Splines allows to create a map that can be evaluated everywhere, and reduces the number of degrees of freedom needed to store the constructed (gradient) map, by using e.g. high order B-Splines functions.