基于投影二分法的车牌字符分割方法  

A License Plate Character Segmentation Method Based on Projection Dichotomy

作　　者：冼允廷[1] 路小波[1] 施毅[1] 钟琨[1] 

机构地区：[1]东南大学,南京210096

出　　处：《交通与计算机》2007年第5期69-72,共4页Computer and Communications

基　　金：高等学校科技创新工程重大项目培育资金项目资助(批准号:705020);江苏省自然科学基金项目资助(批准号:BK2004077)

摘　　要：在车牌字符分割中,传统的投影法对车牌图像质量的要求较高,受噪声影响较大,易造成分割字符的粘连与断裂,在一定程度上影响了车牌的识别率。针对传统投影法的不足,文章提出了一种基于投影二分法的字符分割算法,该方法能快速有效地把车牌字符准确地提取出来。实验结果表明,该方法抗干扰能力强,能有效地抑制字符的粘连,减少字符的断裂,车牌识别准确率高。The traditional projection requires the license plate picture with high quality, and it is seriously subjected to the noise. Moreover, it is easy to induce the segmentation character adhesion and break, affecting the license plate recognition rate in a certain extent. In view of the disadvantage of traditional methods, an algorithm of character segmentation was presented based on the projection dichotomy. This method can segment the license plate character more effectively and accurately, avoid the character adhesion, and reduce the character fracturation. The experimental results show that this method has strong antinoise ability, and that the rate of accuracy is high.

关 键 词：车牌识别 字符分割 二分法 垂直投影 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]