水利工程混凝土大坝水下裂纹检测方法研究  

Research on Underwater Crack Detection Method for Concrete Dams in Water Conservancy Engineering

作　　者：李金花 曾凡妮 Jinhua Li;Fanni Zeng(Shandong Zouping Water Conservancy Engineering Office,Binzhou,Shandong 256200,China)

机构地区：[1]山东邹平水利工程处,山东滨州256200

出　　处：《产业科技创新》2024年第5期91-94,共4页Industrial Technology Innovation

摘　　要：由于现有的检测方法平均精度较低,检测误差大,为此研究水利工程混凝土大坝水下裂纹检测方法具有重要意义。本文运用机器视觉的方式采集水利工程混凝土大坝水下裂纹图像,对复杂图像进行均衡化处理和滤波处理,得到平滑图像来减少噪声。通过图像分割获取不同区域内部的裂纹特征。将图像分解成不同尺度的频子图像,计算中心矩。对不变矩进行处理,构成整个图像的特征矢量完成特征提取。构建水下裂纹边缘检测模型,根据灰度的差异,对裂纹图像进行像素点扫描。根据边缘与基点位置获得更准确的裂纹参数信息完成检测。实验结果表明,随着训练轮数增加,实验组的平均精度达到了90%,获得更加精确的结果;经过10组测试得到的裂纹位置检测误差在0.5~1mm之间,处于实际检测误差范围之内,准确定位裂纹位置,检测效果显著。Due to the low average accuracy and large detection errors of existing detection methods,it is of great significance to study underwater crack detection methods for concrete dams in hydraulic engineering.This article uses machine vision to collect underwater crack images of concrete dams in hydraulic engineering,and applies equalization and filtering to complex images to obtain smooth images to reduce noise.Obtain crack features within different regions through image segmentation.Decompose the image into sub images of different scales and calculate the central moment.Process the invariant moments to form the feature vector of the entire image for feature extraction.Construct an underwater crack edge detection model and scan the crack image pixel by pixel based on the difference in grayscale.Obtain more accurate crack parameter information based on the edge and base point positions to complete the detection.The experimental results showed that as the number of training rounds increased,the average accuracy of the experimental group reached 90%,achieving more accurate results;The crack position detection error obtained through 10 sets of tests is between 0.5 and 1mm,which is within the actual detection error range.The crack position is accurately located and the detection effect is significant.

关 键 词：水利工程 混凝土大坝 水下裂纹检测 灰度差异 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]