Abstract

The principle of Extreme Physical Information (EPI) permits the derivation of physical probability density functions (PDF’s). Consider the flow of Fisher information J→I from object to data space that takes place during a measurement. This obeys the dual effect I–J = extrem., I–κJ = 0, κ = const., whose variational solution is the PDF. An optical object profile o(x) may be regarded as a PDF whose ”derivation” is its restoration. We show how EPI may be applied to numerically restore optical objects in this way.

© 2001 Optical Society of America