Overlapping cell segmentation based on improved maximization of interclass varia

Abstract

The phenomenon of overlapping cells in red blood cell images affects the diagnosis and prevention of diseases. Therefore, an overlapping cell segmentation based on the improved Otsu(Maximization of interclass varia) algorithm is proposed. First, the red cell image is preprocessed, and then the improved Otsu algorithm is used to segment the red cell image to distinguish the red cell region from the background region. Then, based on the feature that the red blood cell image is approximate to a circle, the radius of the red blood cell is obtained by particle size measurement, the contour of the red blood cell is obtained by chain code, and the geometric relationship is combined to obtain the center of the red blood cell. Finally, MSF is used to measure the effect of red blood cell segmentation, and several experiments are carried out to verify the reliability of this method.