基于图像统计的路面磨损状况感知  

Perception of Pavement Wear State Based on Image Statistics

作　　者：张彦 陈华[1] 马少陈 谢承昊 ZHANG Yan;CHEN Hua;MA Shao-chen;XIE Cheng-hao(School of Automobile and Traffic Engineering,Wuhan University of Science and Technology,Wuhan 430081,China)

机构地区：[1]武汉科技大学汽车与交通工程学院,湖北武汉430081

出　　处：《物流工程与管理》2022年第10期80-84,共5页Logistics Engineering and Management

基　　金：国家自然科学基金、国家重大科研仪器研制项目(编号:51827812);基于光谱特征的沥青路面纹理构造和材质的综合表征技术研究(编号:2017xz020)。

摘　　要：路面磨损状态是路面技术状况评定的重要指标之一。在分析图像指标特征和数据分布特征的基础上,以沥青道路路面图像为研究对象,运用三维激光扫描仪采集并计算了路面磨损状态的评价指标MPD,对运用手机拍摄的共488张数据图像进行指标分析,计算了七个常规纹理指标,定义并计算了新的参数——亮暗比,并引入图像构造深度指标。随后,研究了这些纹理指标与路面磨损状态评价指标MPD之间的相关性,基于残差曲线和回归系数表,选定用熵、能量和图像构造深度三个指标构建模型。结果显示,磨损较轻路面回归模型具有很好的相关性,磨损较重路面回归模型的相关性较差,为运用简单图像统计特征分析表示路面磨损状态提供了一定的参考。The state of pavement wear is one of the important indicators for evaluating the technical condition of the pavement.Based on the analysis of the image index characteristics and the data distribution characteristics,this study took the selected pavement image on a certain asphalt road as the research object.A three-dimensional laser scanner was used to collect and calculate the evaluation index MPD of the road wear state,and the index analysis was carried out on a total of 488 data images taken with mobile phones.The study calculated seven conventional texture indicators and a new indicator parameter——light-dark ratio and introduced image structure depth indicator.The correlation between these texture indicators and the evaluation index MPD of the road wear state is studied,and based on the regression coefficient table and residual curve analysis,the model is finally constructed by using three indicators:entropy,energy and MPD.The results show that the regression model of lightly worn pavement has good correlation,and that of heavily worn pavement has poor correlation.This study provides a certain reference for using simple image statistical feature analysis to express the pavement wear state.

关 键 词：道路检测 路面磨损 图像统计 纹理指标 多元回归分析 

分 类 号：U416.2[交通运输工程—道路与铁道工程]