ImWIP: Image warping for Inverse Problems

ImWIP provides efficient, matrix-free and GPU accelerated implementations of image warping operators. The goal of this package is to enable the use of image warping in inverse problems. This requires two extra operations on top of regular image warping: adjoint image warping and differentiated image warping.

Introduction:

• Why ImWIP?
• Installation
• Examples
• Benchmarks
• Citing ImWIP

Reference Documentation:

• API Reference
  • imwip.functions
  • imwip.operators
  • imwip.matrices
  • imwip.solvers

Bibliography:

BFB94

John L Barron, David J Fleet, and Steven S Beauchemin. Performance of optical flow techniques. International journal of computer vision, 12:43–77, 1994.

BB88

Jonathan Barzilai and Jonathan M Borwein. Two-point step size gradient methods. IMA journal of numerical analysis, 8(1):141–148, 1988.

CPH+21

Umberto Celano, Kristof Paredis, ADL Humphris, Matt Tedaldi, C O'Sullivan, JP Hole, and J Goulden. An innovative probe microscopy solution for measuring conductivity profiles in 3-dimensions. In Metrology, Inspection, and Process Control for Semiconductor Manufacturing XXXV, volume 11611, 116110J. International Society for Optics and Photonics, 2021.

CV17

Umberto Celano and Wilfried Vandervorst. Nanoscale three-dimensional characterization with scalpel SPM. Conductive Atomic Force Microscopy: Applications in Nanomaterials, pages 187–210, 2017.

CLC+08

Mingli Chen, Weiguo Lu, Quan Chen, Kenneth J Ruchala, and Gustavo H Olivera. A simple fixed-point approach to invert a deformation field a. Medical physics, 35(1):81–88, 2008.

RSDB21

Jens Renders, Jan Sijbers, and Jan De Beenhouwer. Adjoint image warping using multivariate splines with application to four-dimensional computed tomography. Medical Physics, 48(10):6362–6374, 2021.