基于连通域提取的车牌字符分割算法

License Plate Character Segmentation Based on Connected Component Extraction

车牌字符分割是车牌识别系统的三大关键技术之一。为了准确地进行字符分割,提高字符识别的准确率和识别的速度,提出了一种改进的连通域提取车牌字符分割算法,算法通过扫描目标像素的若干相邻像素后按照一定的准则确定目标像素与相邻像素之间的连通关系,可克服同类算法中像素重复标记和归并标记需大量运算的缺陷,最后得到正确的连通标记划分并能提取出各个分离的连通区域。实验表明算法能够准确快速地识别字符,且具有较好的稳健性。

License plate character segmentation is one of the three key technologies of Vehicle License Plate Recognition System.A good algorithm or method which can segment characters accurately,can improve the accuracy of character recognition and make the identification very fast.This paper presents an improved algorithm for extracting connected areas in binary images,scanning the adjacent pixels of target pixels and determining the connectivity of the relationship between them according to several rules.The algorithm overcomes the shortcomings of needing a memory of big volume to memorize connecting relationships and a great deal of operations to unite labels.The algorithm can ascertain the right distribution and extract the connected region of separation at last.Experiments show that the algorithm can accurately and quickly segment characters and has good robustness.