We consider the problem of filtering a set of images lying on a low dimensional manifold, under the assumption that the in-plane rotation of each image is irrelevant. We derive the steerable graph Laplacian on the image-manifold, which accounts for all planar rotations of all images, and show how to use it for image filtering while exploiting all images and their rotations simultaneously. We demonstrate our approach for the denoising of cryo-electron microscopy image datasets.

This presentation is part of Minisymposium “MS42 - Low dimensional structures in imaging science (3 parts)”
organized by: Wenjing Liao (Georgia Institute of Technology) , Haizhao Yang (Duke University) , Zhizhen Zhao (University of Illinois Urbana-Champaign) .

Authors:

Boris Landa (Tel Aviv University)

Yoel Shkolnisky (Tel Aviv University)

Keywords:

graph laplacian, image reconstruction, image representation, manifold learning