Between-class variance as a measure of image bimodality and sharpness

This paper investigates the suitability of between-class variance as a measure of image bimodality and sharpness. Between-class variance has been first proposed as a criterion function to determine an optimal threshold to segment images into nearly uniform regions. The evaluation on artificial images and real images of human aortic sections showed that normalized between-class variance can be effectively used to detect image bimodality, as well as to evaluate image sharpness. While the former can be utilized in image segmentation, the latter is essential to ensure repeatability in quantitative microscopy