Skip to content

fabiansvara/Alignment_Projects

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Aligner

What it does well

Stitches together any mode of serial section image data:

  • TEM
  • Block-face SEM
  • Fluorescence Array Tomography

Fast and scalable:

  • Runs on linux cluster using Sun Grid Engine API
  • Align 2 to billions of images
  • Approx. linear time/volume & mem/volume scaling
  • Two million 4MB images align in about 8 man-hours

Handled pathologies:

  • Missing tiles or whole sections
  • Fragmented / small / irregular sections
  • Arbitrarily rotated/translated sections
  • Burns, scars, foreign matter
  • Exposure inhomogeneity

Input:

  • 8 or 16 bit TIF, PNG, MRC
  • Simple meta-data as text or TrakEM2 XML

Output:

  • Basically 1 affine or homographic transform per image tile
  • Flexible output as text tables or TrakEM2 XML files

Limitations

  • One linear transform / tile; not an elastic aligner
  • Unfinished handling of geometry-altering folds and tears
  • All images in a data set must be of same fixed dimensions

Requirements

  • Linux cluster with (Sun / Oracle / Univa) Grid Engine.
  • Fiji + TrakEM2: invaluable image and stack tools.

Documentation

  • Project folder 00_DOC details installation and methods
  • Alignment_Tutorial walks you through a real-world example

Authorship

Developed over several years at HHMI/Janelia Research Campus, originally by Louis Scheffer, and subsequently refined into current form by Bill Karsh. See reference "Automated Alignment of Imperfect EM Images for Neural Reconstruction".

About

Aligner for large scale serial section image data

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • C++ 85.9%
  • C 10.8%
  • Shell 1.4%
  • Makefile 1.0%
  • Python 0.7%
  • Assembly 0.1%
  • SourcePawn 0.1%