基于灰度相关性的裂缝检测算法

Crack detection algorithm based on gray correlation

从裂缝的灰度特征入手,提出一种基于灰度值相似度的裂缝提取方法。针对混凝土路面孔洞、污损等因素干扰,主要分三步实现裂缝提取:a)对所采集的路面图像进行特征分析,建立灰度值相似度函数,根据相似度对比提取疑似裂缝信息;b)利用连通域度量去除虚假裂缝,将问题转换为图论问题;c)利用谱聚类算法提取真实裂缝。通过大量混凝土路面图片测试表明,该方法能实现路面真实裂缝特征提取,具有较强实用性。

From the characteristics of the cracks gray,this paper proposed a method of extraction cracks based on gray value similarity to achieve the accurate identification of cracks from the image. For these disturbed factors of pavement holes and de- facements, cracks extraction were performed by the following three steps：a）it analyzed the cracks＇ characteristics of the ac- quired pavement images, built a similarity function of the gray value and extracted suspected information of the cracks according to the similarity comparison;b）it removed the false cracks by the computed measurement of connected domain and transformed the crack detection problem into graph theory one; c）it extracted the true cracks by the spectral clustering algorithm. By tes- ting many images of concrete pavement, the testing results show that the proposed method can extract the pavement cracks＇ characteristics actually and have good practicability.