Comparison of the methods of microscopic image segmentation

Abstract

The goal of the investigation is to examine and to compare results of the microscopic image segmentation methods proposed by Authors. The comparison is a part of investigation, concerning methods of cells culture growth monitoring in the sense of confluence and counting a number of cells in culture. The results of chosen methods of cells’ region detection in a microscopic image are compared in the paper. The first method, the reference method (R), is based on the image reorganization and conventional thresholding. The second, the morphological flattening (MF) method, is based on image morphological filtering and double thresholding. The third, the watershed method (WS), is based on the morphological watershed, done on a simplified and enhanced image. The comparison has been done on the results of cells’ area detection in the bright field microscopy images of the neural stem cells’ culture. The WS method detects cells’ area more precisely than others.