类间方差法对噪音和目标大小十分敏感，它仅对类间方差为单峰的图像产生较好的分割效果。

当目标与背景的大小比例悬殊时，类间方差准则函数可能呈现双峰或多峰，此时效果不好，但是类间方差法是用时最少的。-Variance between class size of the noise and the target is very sensitive, it only to the between class variance to unimodal images produce better segmentation effect.

The disparity between the target and the background when the size ratio, the between class variance criterion function may present Shuangfeng or multi peaks, the effect is not good at this time, but the between class variance method is used when the least.