Skip to main content
. Author manuscript; available in PMC: 2018 Mar 27.
Published in final edited form as: Comput Intell Methods Bioinform Biostat. 2017 Oct 17;10477:42–58. doi: 10.1007/978-3-319-67834-4_4

Algorithm 1.

Automated image registration procedure (Fig 4) to find the least off-center patch from a given commodity camera video frame. The whole slide image is split into N overlapping 800 × 800 px patches. “Three or fewer patches spatially removed” means any In must be (i) Iprior, (ii) adjacent to Iprior, (iii) adjacent to a patch adjacent to Iprior, or (iv) adjacent to a patch adjacent to an Iprior-adjacent patch. In this way, In is restricted to a spatial neighborhood localized around the prior match, typically improving image registration performance because most slide movements are small. On lens change, (i) the patch at lower magnification and (ii) the patches at higher magnification covering the same area as the current magnification’s neighborhood are considered for registration only.

graphic file with name nihms931049f11.jpg