Bibliogr. s. 71

In this paper we present a method to solve a problem of brightness changes within a stack of images obtained by confocal microscope. A result of specimen scanning is a series of 2D images. Consecutive images show deeper sections of stained tissue. Within a stack of images a light attenuation increases with the depth of imaged focal planes. This well known physical phenomenon is a major obstacle that stands in the way of any CLSM image based analysis. Because light attenuation can be compensated by histogram modeling techniques, a straightforward solution is reached by histogram equalization of all images in a stack. The aim is to remodel shapes of histograms in such a way that the intensity histograms of the resulting images become uniform within all possible brightness values. Numerous histogram equalization methods have been developed in the last decade. In this paper we propose a novel 3D histogram equalization algorithm. It is tuned to balance contrast and brightness of images within one stack. In this approach we assume a linear correlation of brightness among adjacent stack slices. Unlike other known algorithms our approach concentrate more on preservation of information than improvement of visual aspects. The presented algorithm was tested on real data as a module embedded in the system for automatic 3D reconstruction of brain glial cells. Because this novel method is a powerful and effective tool for contrast enhancement and achieving evenly balanced luminescence of confocal stack images it substantially contributes to the quality of fully dimensional cell reconstruction.