Automated characterization of bacteria in confocal microscope images

The problem of bacteria characterization from images acquired in microscopy is considered. The proposed approach consists of four consecutive steps: 1) images enhancement; 2) segmentation based on Watershed algorithm; 3) optimization of the initial segmentation; 4) bacteria quantification using a template matching procedure. Additionally, a modified version of this first approach allows to characterize bacteria colonies. Finally, an analysis of results is performed by comparing the estimated results with respect to results obtained manually by the expert.