Rapid Automatic Segmentation of Fluorescent and Phase-Contrast Images of Bacteria

Abstract

Image segmentation, i. e. the division of an image into regions with different properties, is one of the most important steps in image analysis. All results derived from all further steps, such as morphometry or fluorimetry, depend critically on the segmentation algorithm’s ability to determine the exact borders between, e. g. objects and background. In many, if not most cases, image segmentation is performed by manual setting of a threshold. All pixels with grey levels above the threshold are considered part of an object, all others background (or vice-versa, depending on the stain). Though simple to implement and reasonably fast, it is by no means the method of choice, since it is not objective and therefore not very reproducible. Besides, if different objects have been stained with different intensities it is impossible to set a single threshold for all objects within one field of view.