Journal of Pattern Recognition Research

Classic statistical thresholding methods based on maximizing between-class variance and minimizing class variance fail to achieve satisfactory results when segmenting small objects. In this paper, a new thresholding objective function is formulated for segmenting small object based on histogram projection by Fisher discrimination. The proposed method determines the optimal threshold is selected by the Fisher discriminant criterion; namely, by maximizing the measure of separability of the resultant classes in gray levels. The method was compared with several classic thresholding methods on a variety of small medicals images, and the experimental results show the effectiveness of the method.