基于图像特征转换和级联分类器的混凝土表面裂缝快速识别方法  

Study on Fast Recognition Method of Concrete Surface Cracks by Image Feature Transformation and Cascade Classification Model

作　　者：孙瑞猛 刘伟 喻东晓 刘妍妍 SUN Rui-meng;LIU Wei;YU Dong-xiao;LIU Yan-yan(Botou Department ofTransportation,CangZhou 062154,China;Road Transport Business Development Center ofCangZhou,Cangzhou 061014,China;Hebei Xingcang Engineering Testing Technology Co.Ltd.,Cangzhou 061000,China;Hebei University of Water Resources and Electric Engineering,Cangzhou 061001,China)

机构地区：[1]泊头市交通运输局,河北沧州062154 [2]沧州市道路运输事业发展中心,河北沧州061014 [3]河北兴沧工程检测技术有限公司,河北沧州061000 [4]河北水利电力学院,河北沧州061001

出　　处：《工程建设与设计》2023年第21期135-138,共4页Construction & Design for Engineering

基　　金：沧州市重点研发计划项目(213101015);泊头市交通运输局横向课题(2021130901000001);河北省大中学生科技创新能力培育专顶(22E50430D)。

摘　　要：提出了一种基于图像特征转换和级联分类器模型的快速混凝土表面裂缝快速识别方法,并利用交叉验证法对网络结构进行了优化。研究结果表明:将裂缝图案经特征变换后转化为一组特征值,能够有效突出裂缝的形态特征,极大降低运算复杂性;级联分类器模型对样本库容量要求更低,效率较高,具有更高的鲁棒性及识别准确度。A fast recognition method based on the analysis ofimage pattern characteristics and cascade classifier is established.Optimizations of network structures were conducted by cross validation.Results indicate that the eigenvalues transformed by the pattern characteristics highlight the morphological characteristics offractures,simplifying the detection algorithm for concrete surface;The cascaded classifier model has lower requirements on sample base capacity,higher efficiency,higher robustness and recognition accuracy.

关 键 词：混凝土表面 裂缝识别 特征转换 级联分类器 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]