摘要
针对车牌字符分割算法在以往单一分割算法的实现下分割正确率低这一普遍问题,提出了一种改进的8连通域标记算法对车牌进行粗分割,该算法只需要一次扫描就可以完成连通域的标记,可以避免传统连通域算法中像素的重复标记和归并标记的大量运算问题。将粗分割结果与先验车牌结构作比较,再结合垂直投影法进行细分割,最终实现车牌字符的准确分割。实验结果表明,该算法可以有效地解决汉字不连通、字符的断裂与粘连等常见问题。
Considering the low accuracy of the previous single license plate character segmentation algorithm with common problems, an improved 8 connected component labeling algorithm of license plates for coarse segmentation is proposed. The result of this algorithm is very simple, most of the characters can be applied with one scanning. This algorithm overcomes the problem of needing a memory of big volume to memorize connecting relationships and a great deal of operations to unite labels. The results with the known plate structure are compared. Then, adaptive vertical projection algorithm is introduced for further segmentation. License plate character segmentation is implemented finally. The experimental results show that the segmentation method in this paper can effectively solve unconnected problem of Chinese character of fracture and adhesion.
出处
《自动化技术与应用》
2015年第12期93-97,共5页
Techniques of Automation and Applications
关键词
字符分割
连通域
垂直投影
character segmentation
pixel labeling
vertical projection