基于局部纹理特征的沥青路面裂缝检测方法

Asphalt pavement crack detection method based on local texture features

针对不均匀光照和阴影等因素影响沥青路面图像中裂缝检测误识别问题,提出一种基于局部纹理特征的沥青路面裂缝检测方法。设计结构保持型Retinex算法将高频的纹理信号从低频光照信号和结构型纹理中分离,改进百分比阈值算法,获取高信噪比的裂缝区域二值图像,建立高置信裂缝段的特征匹配机制,利用圆形度、面积、不同置信裂缝的类间欧氏距离进行置信连通域邻域去噪,实现裂缝区域的提取。实验结果表明,该算法改善了不均匀光照和阴影情况下裂缝识别的准确性。

Aiming at the problem that uneven illumination and shadow affect the recognition of cracks in asphalt pavement image,a method of asphalt pavement crack detection based on local texture features was proposed.The structure preserving Retinex algorithm was designed to separate the high-frequency texture signal from the low-frequency illumination signal and the structural texture,and the percentage threshold algorithm was improved to obtain the binary image of the fracture region with high signal-to-noise ratio.The feature matching mechanism of high confidence fracture segment was established,and the noise of confidence connected region neighborhood was removed using circularity,area and Euclidean distance between different confidence cracks to extract the fracture area.Experimental results show that the proposed method improves the accuracy of crack recognition under the condition of uneven illumination and shadow.