Measurements within the space of images are necessary for many problems in computer vision. Geometry goes beyond measurements and considers the symmetries of the space. We understand images as maps between smooth manifolds and consider Lie groups acting on the domain and codomain. We constrain distances to be invariant to the action and compute the discrepancy: the energy to deform one image to another through the group action.
|Qualification||Doctor of Philosophy|
|Award date||23 Oct 2017|
|Publication status||Unpublished - 2017|