Fast regularization technique for expectation maximization algorithm for optical sectioning microscopy

Publisher Website
Google Scholar
Abstract
Maximum likelihood image restoration is a powerful method for 3D computational optical sectioning microscopy of extended objects. With punctate specimens, however, this method produces a few very bright isolated spots and dim detail around them is lost. The commonly used regularization methods (sieves and roughness penalty) decrease the amplitude of the bright spots, but do not avoid loosing dim detail. We derived an intensity regularization that decreases the amplitude of bright spots without loosing dim detail. In contrast with other regularization methods, this method does not increase significantly the computational complexity of the estimation algorithm.

This publication has 0 references indexed in Scilit: