论文部分内容阅读
建筑物墙体裂纹是重要的安全隐患,检测混凝土墙体表面的裂纹及测量其最大宽度,已引起众多关注.现介绍基于图像处理的智能检测方法,即根据裂纹像素点分布特征,利用连通域面积大小来提取裂纹,并删除伪裂纹等杂质,再对含有分支或网状裂纹进行局部处理,根据裂纹特征像素点的位置关系获取聚类近似初始值,之后利用K-means聚类算法不断迭代计算裂纹特征像素点到其对应直线的最短距离,并以此将图像中的像素点归为不同方向的裂纹类.最后,利用分类好的裂纹像素点分别进行边缘检测与最大宽度测量并比较,来获取含有交叉裂纹的最大宽度值.本文获得的水平裂纹最大宽度的相对误差为2.968%,斜垂裂纹最大宽度的相对误差为5.188%.
Building wall cracks is an important security risk, detection of concrete wall surface cracks and measuring the maximum width, has attracted a lot of attention.An image processing based intelligent detection method is introduced, that is based on the distribution of crack pixel features, the use of connectivity Area size to extract the crack, and remove the pseudo-crack and other impurities, and then containing the branches or reticular crack local processing, according to the crack location of the relationship between the location of pixels to obtain approximate clustering initial value, and then use K-means clustering algorithm iterative The shortest distance between the crack feature pixel and its corresponding line is calculated and the pixels in the image are classified as cracks in different directions.Finally, the edge detection and the maximum width measurement are compared and compared respectively with the classified crack pixel, To obtain the maximum width of the cross-cracks.The relative error of the maximum width of the horizontal crack obtained in this paper is 2.968% and the relative error of the maximum width of the slash-crack is 5.188%.