一种车牌号码图像二值化的新方法  

New Method for Vehicle License Plate Image Binarization

作　　者：潘梅森[1] 张奋[1] 雷超阳[2] 

机构地区：[1]湖南文理学院计算机科学与技术系,常德415000 [2]湖南长沙通信职业技术学院,长沙410000

出　　处：《计算机工程》2008年第4期209-211,214,共4页Computer Engineering

基　　金：湖南省自然科学基金资助项目(05JJ40098);湖南省教育厅科研基金资助项目(05C720)

摘　　要：利用数学形态学的开运算消除图像光照不均匀影响,根据欧几里德距离确定图像阈值。基于人类视觉特性确定每个像素点的视觉敏感度系数,并根据图像像素点灰度值、图像阈值和视觉敏感度系数对图像进行二值化处理。该方法能有效地消除伪影,较好地保护字符笔画。模拟实验结果表明,该方法是有效的。This paper eliminates the image non-uniform brightness using mathematical morphological open operation and obtains the image threshold value according to the Euclidean distance. Standard deviation is used to determine the visual sensitivity coefficient of each pixel according to the human visual characteristics. Image binarization is performed according to the gray value, the image threshold and the visual sensitivity coefficient. This method both can effectively eliminate the ghosts, and can keep the character stroke well. Experimental results show that this method is more effective.

关 键 词：二值化 阈值 人类视觉系统 像素 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]