摘要
文档图像的二值化是光学字符识别(OCR)的基础,本文在实验的基础上通过对现有的二值化算法进行研究分析,综合比较了期望灰度法、Otsu方法、迭代最优方法、Niblack方法、平均梯度法和四叉树分解方法,分析了几种算法的优缺点,并对其发展趋势进行了简要的论述。
The binarization is the key problem in OCR. In this paper, the existing methods are analyzed, including expected gray value method, otsu method, recursive optimal method and division based method. A comprehensive comparision based on the experiments is carried out to show their advantages and drawbacks, the developing trends is also discussed briefly.
出处
《中国传媒大学学报（自然科学版）》
2008年第4期66-70,共5页
Journal of Communication University of China：Science and Technology
