【摘要】 一维Otsu法是一个经典的阈值分割方法,遵循该方法的构造思想,Li与Lee基于交叉熵提出了一个阈值分割方法.在解释和说明Li与Lee的方法的基础上,与二维Otsu直线型阈值分割方法相对应,提出二维灰度直方图上的二维交叉熵直线型阈值分割方法并给出快速递推公式.与二维Otsu直线型阈值分割法相比,新方法能够更好的适应目标和背景方差相差较大的情形,是一个有效的阈值分割方法.更多还原

【Abstract】 One-dimensional Otsu’s method is a classical threshold segmentation method.Following this thought,Li and Lee proposed a threshold segmentation method based on cross-entropy.A two-dimensional cross-entropy linear-type threshold segmentation method and its fast recursive formula were proposed based on the explanation and demonstration of Li and Lee’s method and corresponding with the two-dimensional Otsu’s linear-type threshold segmentation method.Compared with the two-dimensional Otsu’s linear-type threshold segmentation method,the new method can be better adapted to the cases that the variance between the object and background is large.It is shown that the new method is an available threshold selection method.更多还原