Analytical registration of vertical image drifts in parallel beam tomographic data

Abstract

Reconstructing tomographic images of high resolution, as in x-ray microscopy or transmission electron microscopy, is often limited by the stability of the stages or sample drifts, which requires an image alignment prior to reconstruction. Feature-based image registration is routinely used to align images, but this technique relies on strong features in the sample or the application of gold tracer particles, for example. In this Letter, we present an analytic approach for achieving the vertical registration based on the inherent properties of the data acquired for tomographic reconstruction. It is computationally cheap to implement and can be easily integrated into existing reconstruction pipelines.