Otsu是按照图像的灰度特性，将图像分成目标和背景两部分，背景和目标之间的类间方差越大，说明构成图像的两部分的差别越大。当部分目标错分成背景或者部分背景错分成目标，都会导致两部分的差别变小。因此，使类间方差最大的分割意味着错分概率最小。-Otsu is based on gray-scale image features, the image is divided into two parts, the target and background, background and objectives of the greater between-class variance, indicating the composition of the image the greater the difference between the two parts. As part of the target into the background or the wrong part of the background into the wrong target, the difference between the two parts will lead to smaller. Therefore, the maximum between-class variance of the probability of misclassification means that the smallest division.