Alternative Image Stacking Methods

[garrettwrong]

In other low SNR scientific imaging domains it is commonplace to see a variety of image stacking techniques available. The most common of which I've outlined below.

• Mean
  • Theoretically lowest total noise across entire image. Currently what we use.
• Median
  • Outlier Rejection
• Kappa Sigma
  • Computes variance per pixel and rejects pixel contribution if more than sigma deviations.

Additionally there are alternatives based on various weighting statistics, adaptive techniques, and other ideas relating to entropy.

From the software side, it is possible, perhaps even straightforward, to implement such things in ASPIRE. However, more work would be required to explore what is the current state of the art and to confirm/deny improvement of cryo-em data specifically.