【作者】 陆健；

【导师】 程建川；

【作者基本信息】 东南大学 ， 道路与铁道工程， 2006， 硕士

【摘要】 路面破损检测是道路养护管理工作的重要内容之一。传统的人工处理方法速度慢、危险、影响交通、不精确,已不能适应路面破损检测的需求。随着光电技术及计算机技术的快速发展,无损检测技术应用于路面破损检测是必然趋势。本文研究了目前先进的路面图像系统在路面破损自动检测中的应用。它利用高速度、高精度摄像机快速地拍摄路面图像,通过计算机快速处理,得到破损信息。然而,路面图像的复杂性、多样性以及破损信息的弱信号性,使得研究高效的检测算法并不容易。基于种情况,本文深入研究了数字图像处理技术的各种算法,包括:尝试综合多种图像增强以及灰度图像处理技术对路面图像进行适当的预处理;图像二值化以及二值化处理的方法;回顾了经典的边缘检测算法。在此基础上,结合大量路面图像的反复试验得到了路面图像自动处理的有效算法组。同时,本文还系统研究了精度更高的基于BP神经网络的黑箱预测方法,给出了一个基于BP网络的试验平台。最后,本文利用面向对象程序设计方法,设计了实验软件,实现了以上所述的部分算法,并取得了令人满意的效果。整个软件系统是在.NET平台下用C#语言编写而成,界面友好,易于操作,可以实现对沥青路面图像的预处理、图像分割和相关的特征计算,还可观看每步的处理效果和图像的直方图分析。本系统可望应用于对沥青路面破损进行自动检测,为公路管理、养护部门的决策提供科学依据。更多还原

【Abstract】 Pavement distress detection plays an important role in the highway management and maintenance. The traditional manual methods can’t meet the needs of pavement distress detection now, because they have the following shortcomings such as time-consuming, non-precision, dangerous, costly and also affecting transportation. As the quick expend of computer and photoelectricity technology, the application of road non-destructive test for road pavement distress test is necessary.This paper investigates the application of the advanced digital image system in asphalt expressway pavement test, which uses high rapid and precise cameras to screen pavement images, with fast processing of computers, to get distress information. However, because of highway pavement images’ complexity and diversity, and also distress information’s weak-signal, it is not easy to get efficient detection algorithms.In this paper, lots of algorithm of digital image processing is analyzed such as the efficient pre-processing algorithm of highway pavement image which integrates manifold technologies of image enhancement and gray scale image processing, the ways of binarizing image and binary image processing, a large number of classical algorithms of edge detection. Upon the repeating tests of vast pavement images, the author gets a series of effective algorithms for analyzing pavement distress automatically. BP artificial neural network is also analyzed in this paper, which provides a more accurate way of digital image processing, and gets a platform based on BP artificial neural network.Finally, the author has compiled a set of software with OOP (Object Oriented Program) method, which is compiled by C# on the platform of .NET. Using the software, we have checked up the algorithm. Applying this software, the user can look on every image processed step by step and the gray diagram analysis of the image. In addition, the system in this research can be applied to analyze asphalt pavement distress automatically, and provide scientific ground for the highway management and maintenance department.更多还原