PDF全文阅读

张建勋

（平顶山龙泽公路工程有限公司 ，平顶山 467000)

摘   要 ：针对冻融裂缝特征收敛效果差，路面冻融裂缝检测精准度低的问题，设计了基于特征图的路面冻融裂缝红外检测算法。基于红外热成像原理，联合异性温度场重构和双边滤波理论方法，形成新的路面冻融裂缝红外热图像温度场重构方法，构建冻融裂缝特征矩阵，将路面冻融裂缝红外图像作为输入，优化卷积神经网络的卷积核权值 ，构建基于特征图的检测方法，获取最佳路面冻融裂缝红外成像检测结果。结果表明：该算法温度场重构后可使路面冻融裂缝红外图像更加清晰，并保留图像的边缘信息和轮廓，且在不同类型噪声干扰下， 温度场重构后图像始终具有较高的峰值信噪比。该方法可检测到了不同类型的路面冻融裂缝，可以提升道路安  全运维效果。

关键词：路面冻融裂缝 ；卷积神经网络 ；红外成像 ；热图像消噪 ；裂缝检测 ；机器学习

中图分类号：U416. 2           

文献标志码：A            

文章编号： 1005- 8249   (2025)  02- 0164- 05 

DOI:10. 19860/j.cnki.issn1005 - 8249.2025 .02.028

 

ZHANG Jianxun

(Pingdingshan Longze Highway Engineering Co . ,  Ltd . ,  Pingdingshan 467000 ,  China)

Abstract: Aiming at the problem of poor convergence effect of freeze-thaw crack characteristics and low accuracy of pavement freeze-thaw crack detection, an infrared detection algorithm for pavement freeze-thaw cracks based on feature maps was designed. Based on the principle of infrared thermal imaging, a new method for reconstructing the temperature field of pavement freeze-thaw crack infrared thermal images is formed by combining anisotropic temperature field reconstruction and bilateral filtering theory. A freeze-thaw crack feature matrix is constructed, and the pavement freeze-thaw crack infrared image is used as input to optimize the convolution kernel weights of the convolutional neural network. A feature map based detection method is constructed to obtain the best pavement freeze-thaw crack infrared imaging detection results. The results show that the algorithm can make the infrared images of road freeze-thaw cracks clearer after temperature field reconstruction, and preserve the edge information and contour of the image. Moreover, under different types of noise interference, the reconstructed temperature field image always has a high peak signal-to-noise ratio. The conclusion is that this method has detected different types of pavement freeze-thaw cracks, which can improve the effectiveness of road safety operation and maintenance

Keywords: road freeze-thaw cracks; convolutional neural network; infrared imaging; thermal image denoising; crack detection; machine learning

作者简介： 张建勋  (1973—) , 男，本科， 高级工程师，研究方向：道路 、桥梁 、交通工程等。

收稿日期：2023- 09-04