路面裂缝影像几何特征提取算法

An algorithm to image-based pavement cracks geometry features extraction

传统的基于人工视觉野外检测路面裂缝的方法愈来愈不能适应高速公路发展的要求,其耗人力、耗时、影响交通、危险、测量结果不一致性等缺点要求路面调查自动完成。本文提出一种基于数字影像的路面裂缝自动检测方法:首先对降质线阵扫描路面影像进行增强处理.再基于多尺度空间模型进行影像分割,进而得到裂缝区域的边缘矢量及其几何特征,最终为路面裂缝分类提供有用的依据。实验结果表明,该方法能高效地检测出路面影像中的细小裂缝区域及其几何形态参数。

Conventional human-visual and manual pavement cracks detection method and approaches are not fit to the development of the highway any more, and they possess various drawbacks such as labor-intensive, time-consuming, traffic-influencing, dangerous and result-inconsistent which requires to survey the pavement condition automatically. This paper describes an automatic pavement cracks detection method based on digitized image process, it enhances the line-scan pavement image by using a fast correct method in the first, then processes image segmentation based on multi-scale space model, in the end extracts the geometry features of the pavement cracks which provide useful evidence to pavement cracks classification. Experiment results demonstrate the algorithm can correctly discover tiny cracks even from noise pavement images and can extract geometrical parameters exactly.