基于图像处理技术的城市路面裂缝病害检测

发布时间：2018-05-18 01:16

本文选题：路面病害检测 + 初始分类　；参考：《辽宁工程技术大学》2015年硕士论文

【摘要】：随着我国国民经济的飞速发展,车辆行驶数量和道路里程数不断增长,路面承载压力越来越大,道路的养护任务也越来越繁重,其中,路面病害检测工作是道路养护的重要依据,尤其是城市路面的病害检测工作受到越来越多的重视。在各种城市路面病害类型中,裂缝类病害是一种早期破损形式,一旦不及时修补,将会发展成其它更严重的路面病害类型,因此,如何快速、准确地自动识别城市路面裂缝病害的问题亟待解决。本文围绕城市路面裂缝病害检测方法进行了研究。为了高效识别大量路面图像中的裂缝类病害路面图像,并对其进行破损程度评估,本文提出对路面病害检测工作采取“先定性分析后定量分析”的处理流程。首先,对路面图像进行初始分类,主要包括：利用自适应平滑滤波和Wallis滤波进行路面图像增强、P-tile图像分割、提取路面图像的形状特征参数、利用BP神经网络对路面图像进行分类;然后,对初始分类获得的裂缝类病害路面图像采用多尺度、多方位结构元素形态学和最大熵图像分割的算法进行裂缝提取,并对分割后的二值裂缝图像进一步精细化处理,去除伪裂缝；最后,提取二值裂缝图像的投影特征向量,根据不同类型裂缝的投影特征值差异进行裂缝识别,并且分别统计网型裂缝的外接矩形面积和线型裂缝的总长度,计算路面破损状况指数。实验证明：该方法可以提高路面病害检测的处理效率,实现路面裂缝的准确提取,为道路养护提供有效决策信息。
[Abstract]:With the rapid development of our national economy, the number of vehicles and the number of road mileage are increasing, the bearing pressure of the road surface is increasing, and the maintenance task of the road is becoming more and more heavy, among which, Pavement disease detection is an important basis for road maintenance, especially for urban pavement disease detection. Among all kinds of urban road surface diseases, crack diseases are an early form of damage. If they are not repaired in time, they will develop into other more serious types of pavement diseases. Therefore, how to quickly, The problem of accurate automatic identification of urban pavement crack disease needs to be solved. In this paper, the detection method of urban pavement crack disease is studied. In order to identify the pavement images of crack diseases in a large number of pavement images efficiently and evaluate the damage degree of pavement images, this paper puts forward the processing flow of "qualitative analysis then quantitative analysis" for pavement disease detection. Firstly, the initial classification of road surface image is carried out, including: adaptive smoothing filter and Wallis filter are used to segment the pavement image and extract the shape characteristic parameters of the road image. BP neural network is used to classify the pavement image, and then, the original classification of cracked road surface image is based on multi-scale, multi-orientation structure element morphology and maximum entropy image segmentation algorithm for crack extraction. Finally, the projection feature vector of binary crack image is extracted and identified according to the difference of projection eigenvalue of different types of cracks. The external rectangular area and the total length of the linear crack are calculated respectively and the pavement damage index is calculated. Experimental results show that this method can improve the efficiency of pavement disease detection, realize accurate extraction of pavement cracks, and provide effective decision information for road maintenance.
【学位授予单位】：辽宁工程技术大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U418.6;TP391.41

【参考文献】