Document Actions
How statistics can help you see in the dark!
Guy Nason
Venue:
Lancaster University, Fylde SCR
2004-11-25 14:00
Haar-Fisz transforms are a new class of multiscale variable transforms that can Gaussianize and variance-stabilize intensity sequences. Haar-Fisz transforms for Poisson intensity data perform better than Anscombe and are also usually better than specialist Bayesian algorithms. Haar-Fisz is also extremely fast and simple. We explain the Haar-Fisz transform, demonstrate its performance and demonstrate some low-light image enhancement applications.
