为解决桥梁底部裂缝检测结果受噪声影响大的问题,提出一种基于Bilateral-Frangi滤波的裂缝检测算法。常规Frangi滤波算法的物理模型与裂缝的灰度值分布具有极高的匹配度,但其预处理环节在平滑噪声的同时会去除裂缝的高频信息,即会破坏裂缝的边缘结构。针对此问题,本文优化了常规Frangi算法的物理模型,在预处理环节加入灰度值域的高斯核函数,与原有的空域高斯核函数构成Bilateral高斯核函数,实现了在保边去噪的同时增强裂缝的功能,提高了裂缝检测的准确性。实验结果显示,本文算法的误检率为0.07%,漏检率为3.31%,在对比算法中表现最出色,为高噪声图像的裂缝检测提供了一种新思路。

Bridge crack detection is greatly affected by noise. To solve this issue, this study proposes a new detection algorithm based on the Bilateral-Frangi filter. The conventional Frangi filter algorithm model fits the gray-value distribution of the cracks very well, but as its preprocessing step decreases the high-frequency information of the cracks in noise smoothing, it destroys the edge structures of the cracks. To solve this problem, the physical model of the conventional Frangi filter algorithm is optimized by adding a gray-value-domain Gaussian kernel function in the preprocessing step to construct a Bilateral Gaussian kernel function combined with an original spatial-domain Gaussian kernel function. This algorithm enhances the crack structure and realizes edge protection and denoising at the same time, which improves the accuracy of crack detection. Experimental results show a false detection rate of 0.07% and a missed detection rate of 3.31%. The proposed algorithm provides a solution for crack detection with the best performance in comparison to other algorithms.