摘要：

在利用数字图像技术检测路面裂缝时,由于部分裂缝过窄或被阴影遮挡或被灰尘填充,导致检测出的裂缝目标不连续,严重影响后续的裂缝参数测量和评价。为此,提出一种基于Prim最小生成树的路面裂缝连接算法。利用屋脊边缘检测方法识别所有的可疑裂缝目标,运用裂缝形状特征去除斑点或块状噪声,实现裂缝的粗定位。在此基础上,通过形态学方法提取粗定位裂缝片段的端点,利用Prim算法构造最小生成树实现路面裂缝片段端点的连接,同时使用裂缝的方向和对比度特征去除连接中的强制伪连接;在连接的基础上对裂缝进行填充和增强,得到完整的裂缝分割目标。对200幅路面图像进行算法测试,应用Hausdorff距离对多种算法的分割性能进行评估,实验结果表明,该算法能明显提高裂缝检测目标的连续性,其检测准确率比灰度直方图等算法高出6个~13个百分点。

关键词: 交通工程, 路面养护, 裂缝检测, Prim最小生成树, 裂缝连接, Hausdorff距离

Abstract:

When the digital image processing technology is used to detect the cracks on the pavement,it is very hard to detect an intact structure for the cracks because parts of the cracks are very narrow,or shadowed by other objects,or filled with dust.These seriously affects the accuracy of the crack parameter measurement and damage index evaluation.Aiming at the problems above,a pavement crack connection algorithm using Prim minimum spanning tree is proposed.The ridge detection method is used to mark out all the suspicious cracks targets,with the shape features of cracks to remove the noises like spots or blocks.So all the long or obvious cracks are remained.Using the morphology method,the endpoints of the remained crack segments are extracted,and the Prim algorithm is used to construct a minimum spanning tree and makes all the discontinuous cracks connected.All the forced pseudo connections are deleted through the orientation and contrast characteristics of the cracks.On the basis of connection,the cracks are enhanced by filling operation and an intact crack structure is acquired.200 pavement images with cracks are tested,and the Hausdorff distance is used to evaluate the performance of various algorithms.Experimental results show that the proposed algorithm significantly improves the continuity of the detected crack targets,the detection accuracy rate of which is higher than other algorithms with by 6~13 percentage.

Key words: traffic engineering, pavement mainteance, crack detection, Prim minimum spanning tree, crack connection, Hausdorff distance

中图分类号: 

• TP751.1