大津阈值分割法，也称最大类间方差法，利用图像的灰度分布密度函数定义图像的交叉熵。

交叉熵可看作是两个概率系统（即图像背景及目标）的信息量之间的差异。求出的交叉熵越大，则分割效果越好。

-Otsu threshold segmentation method, also known as Otsu method, the use of gray-scale image distribution density function definition images of cross-entropy. Cross-entropy can be seen as both the probability of system (that is, the image background and objectives) the difference between the amount of information. Obtained the greater the cross-entropy, then partition the better.