The analysis of the properties of the threshold in the Otsu segmentation method can provide theoretical help for improving and applying the Otsu method.This paper proves a conclusion that the threshold of the Otsu method is the average of the means of two classes partitioned by this threshold.Thus

when the difference of the two within-class variances is large

the threshold of the Otsu method tends to be closer to the class with larger within-class variance

which means that more pixels of this class will be classified into the another class.To overcome this problem

this paper presents an improved Otsu algorithm by constraining the search range of the ideal segmentation threshold.Experimental results show the superiority of the proposed algorithm by yielding more reasonable segmentation results compared to the traditional Otsu method.