基于分裂合并算法的道路标线分割方法研究  

Road marking segmentation method for research based on splitting and merging algorithm

作　　者：王健[1] 王润民[2] 惠飞[1] 赵祥模[1] 杜枫[3] 

机构地区：[1]长安大学信息工程学院,西安710064 [2]长安大学现代工程训练中心,西安710064 [3]中国机械工业集团有限公司综合管理部,北京100083

出　　处：《计算机应用研究》2014年第12期3880-3884,共5页Application Research of Computers

基　　金：国家自然科学基金资助项目(51278058;60902075);国家物联网重大示范工程科研项目(2012-364-812-105);中央高校基本科研业务费专项资金资助项目(2013G5240009)

摘　　要：针对现有的道路标线分割方法在阴影、光照、裂缝等干扰下存在分割精度不足的问题,提出了基于分裂合并算法的道路标线分割方法。该方法通过分析道路标线图像的灰度、边缘、纹理特征,首先采用Bresenham算法进行道路标线有无判定,并根据道路标线边缘实现路面检测图像标线区域和非标线区域的分裂,然后采用基于重心连线灰度值的区域合并和基于灰度与直线边缘特征的区域合并两种方法进行道路标线区域合并,最终完成道路标线的准确分割。实验表明,该方法分割精度较高,具有良好的抗噪性能,能够很好地满足道路检测车采集图像实时处理的需求。For the disturbance such as shadows,lights,cracks and other factors,current road marking segmentation methods could not achieve accurate segmentation. This paper proposed the road marking segmentation method based on splitting and merging algorithm. By analyzing the features of gray,edge,texture of road marking lines,it used the bresenham algorithm to judge whether there were road marking lines in the detected images. Road images were divided into road marking areas and non-road marking areas by road marking lines. Secondly,it used two methods which were the region merging method based on the center of gravity lines of gray value and the region merging method based on the gray and edge feature of road marking areas to complete the mergence of road marking areas. The experimental results show that the proposed method,which has a higher accuracy of segmentation and good robustness,can well meet the needs of the real-time processing of images acquired by road detection vehicles.

关 键 词：路面破损检测 道路标线 图像分割 BRESENHAM算法 重心连线 边缘特征 灰度特征 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]