市政道路护栏立柱焊接节点缺陷深度精准检测OA
Accurate detection of welding node defects in municipal road guardrail columns
市政道路护栏立柱角焊缝焊接缺陷深度不一,常规红外检测方法会因热扩散效应,使不同缺陷程度的焊接节点温差信号趋同,导致检测结果产生误差,进而误判结构安全性,给市政道路的安全运行带来潜在风险.因此,通过红外热图像完成市政道路护栏立柱焊接节点缺陷深度检测.利用多尺度引导滤波和局部显著图权重融合增强图像,量化热异常显著性后映射为加权图,用高斯拟合亮度模型等获取分割结果确定缺陷位置范围,再基于区域生长准则和角焊缝坡口方向,实现缺陷深度的精准识别.试验结果表明:所提方法可有效、精准、全面地分割焊接节点缺陷区域,且对500张红外热图像的缺陷深度检测准确率达99%.说明所提方法可精准高效地完成市政道路护栏立柱焊接节点缺陷深度检测,保证市政道路的安全.
The depths of fillet weld defects in municipal road guardrail posts vary considerably.Conventional infrared detection methods are limited by the heat diffusion effect,which causes the temperature difference signals of welding joints with different defect depths to converge,resulting in detection errors and potential misjudgment of structural safety,thereby posing risks to the safe operation of municipal roads.Therefore,this paper proposes a defect depth detection method for welding joints of municipal road guardrail posts based on infrared thermal images.Multi-scale guided filtering and local saliency map weighted fusion were employed to enhance the images.After quantifying the thermal anomaly saliency,the images were mapped into a weighted graph,and a Gaussian fitting luminance model was applied to obtain segmentation results for determining the defect location and extent.Subsequently,based on the region growing criterion and the fillet weld groove direction,precise identification of defect depth was achieved.The experimental results show that the proposed method can effectively,accurately,and comprehensively segment the defect areas of welding joints,and the defect depth detection accuracy reaches 99%for 500 infrared thermal images.The results demonstrate that the proposed method can accurately and efficiently complete the defect depth detection of welding joints in municipal road guardrail posts,ensuring the safety of municipal roads.
李喜玲
郑州市政集团有限公司 河南郑州 450000
市政道路护栏立柱焊接节点角焊缝红外热图像区域生长准则缺陷深度识别
municipal roadsguardrail pillarswelding nodesfillet weldinfrared thermographyregional growth criteriadefect depth identification
《金属加工(热加工)》 2026 (10)
75-83,9
评论