基于自然段落的车牌图像二值化算法  

A new binarization algorithm of license plate image based on nature paragraph

作　　者：王聪丽[1] 侯著荣[1] 王凯[1] 陈鹏[2] 

机构地区：[1]军械工程学院,河北石家庄050000 [2]武警石家庄指挥学院,河北石家庄050000

出　　处：《信息化纵横》2009年第12期58-61,共4页

摘　　要：在车牌图像的反转灰度直方图的基础上,定义了全峰和半峰的概念。利用自然段落的思想,根据自然段落的分布和长度,确定出阈值所在全峰的范围。在此范围内求极值,将求阈值的灰度范围大大减小,降低了图像二值化的算法复杂度和运算量。实验证明,该算法效果良好,即使对于偏亮和对比度低的图像,也能达到满意的效果。Based on the inversed-histogram and natural phrase ,this paper gives two concepts ：full-peak and half-peak ,and then advances a new method of binarization .This method confirms the range of threshold and finds the extremum in it according to natural phrase＇s distribution and length, so it makes the gray range minished and reduces image binarization＇s complexity and operation quantity very much. It proves that this algorithm can make good effect even about low-contrast and very bright image.

关 键 词：反转直方图 自然段落 全峰 半峰 阈值 

分 类 号：TN911.73[电子电信—通信与信息系统]