基于轮廓特征的水下混凝土结构表面裂缝检测算法

Underwater Concrete Structure Surface Crack Detection Algorithm Based on Contour Features

摘要:

针对水下混凝土结构表面裂缝难以准确检测的问题,提出了一种基于轮廓特征的水下混凝土结构表面裂缝检测算法;在对采集到的水下混凝土结构表面图像进行预处理和边缘检测后,首先通过计算图像中边缘的长度和曲率来进行初步判别,区分杂点和部分非裂缝；然后根据裂缝轮廓相互匹配的特征,利用模板搜索来检测出真实裂缝；最后连接裂缝边缘断开部分,优化裂缝检测结果;实验结果表明,与其它算法相比,该算法检测出裂缝的准确率与完整率至少提高了3.54%。

Abstract:

A new crack detection algorithm based on contour features is proposed aiming at the problem of accurate detecting the underwater concrete structure surface crack. After preprocessing and edge detection to the underwater concrete structure surface image, calculating the length and curvature of the edges to distinguish the noise points and part of non-crack. Then according to the match features of the crack, template search is used to detect the real cracks. Finally, connecting the broken parts of the crack to optimize the test results. Experiment results show that the cracks of the proposed detection method are more complete and accurate.