Image Stabilization with ImageJ
How to correct for motion in time-series 2D and 3D images

Sample data

Download: sample.tif

Original TurboReg Descriptor Correct2D MoCo

I find TurboReg/StackReg to be reliably superior to other ImageJ alignment plugins.

2D Image Stabilization

TurboReg is an ImageJ plugin for aligning images using a subpixel registration algorithm described by Thévenaz et al., 1998. StackReg is an ImageJ plugin that applies TurboReg alignment to stacks.

Download:

Launch: Plugins > StackReg

Multichannel images can be converted to RGB, stabilized, then converted back to multi-channel composite images.

3D Image Stabilization

3D image alignment is slower and more complex, so ensure it is truly needed for your application before proceeding.

  • If your goal is to analyze projection images, I recommend projecting 3D times-series data to create a 2D stack then use StackReg to align it.

  • If your goal is to perform analyses at distinct Z positions over time, 3D alignment is required.

Correct 3D Drift is an ImageJ plugin for 3D time-series alignment. It uses a phase correlation based algorithm described by Preibisch et al., 2009. This plugin ships with FIJI and does not have to be manually installed. This plugin can be used on 2D images, but I find its performance inferior to TurboReg/StackReg.

Download: Correct_3D_Drift

Launch: Plugins > Registration > Correct 3D Drift

Additional Resources

The strategies above are those I typically recommend, but additional tools are available for 2D and 3D image alignment:

  • Descriptor Based Series Registration - An ImageJ plugin (provided with Fiji) which may have superior sub-pixel alignment performance but I find it produces output images which are too blurry for my applications.

  • Motion Correlation (MoCo) - An ImageJ plugin based on Dubbs et al., 2016 optimized for rodent brain imaging. This project does not appear to be maintained, and the repository is poorly documented. I find performance to be inferior to TurboReg.