摘要
以门牌号为对象,利用现代图像处理技术,研究移动机器人门牌号识别算法.将彩色图像转换成R G图像,利用正交投影获得了门牌号定位.基于大津阈值分割方法,提出了图像二值化阈值选择算法.针对细化后的二值图像,综合利用端点、叉点、欧拉数和端点斜率,提出了门牌号识别算法.仿真结果证明了所提算法的有效性.
This paper, taking a doorplate as the landmark, studies the doorplate recognition algorithm by making full use of the modern image processing technology. Firstly, a color image is transformed into R-G one, and an algorithm to locate the doorplate is put forward in this paper by quadrature projection. Secondly, an algorithm based on the maximum variance between clusters is obtained, used to select the threshold for the gray image to binary one. Thirdly, aiming at the properties of the thinned characters in the binary image such as end point, cross point, genus and end point slope, an algorithm to recognize doorplate numbers is presented. At last, simulation results prove the effectiveness of the presented algorithms.
出处
《河南大学学报（自然科学版）》
CAS
2004年第2期69-72,共4页
Journal of Henan University:Natural Science
基金
河南省高校青年骨干教师资助计划项目
关键词
移动机器人
定位
分割
细化
识别
mobile robot
location
segmentation
thinning
recognition
