基于分形路面破损图像裂纹识别研究

发布时间：2018-06-29 06:41

本文选题：裂纹检测 + 分形　；参考：《长安大学》2014年硕士论文

【摘要】：目前国内外道路的养护与管理工作主要根据人工检测和自动检测而来的数据信息来实施。因此国内外也都投入巨大费用用于研发各种路面自动采集设备与数据处理系统，一般都是应用多功能道路检测车系统获取路面破损图像、车辙、平整度以及道路周围的环境信息，后续对采集的数据进行分析与识别处理，从而提供养护与维修报告。路面破损数据的信息处理由于受到路面自身的复杂性以及数据在采集过程中所遇到的外部环境影响，使得路面破损图像的数据处理工作量巨大而花费时间很长。因此关于路面破损的图像自动识别系统研究意义重大。本课题研究的内容是对设备采集的路面破损图像实现自动识别，主要是对路面破损图像中的各种裂纹进行自动识别。为了很好识别出破损图像中的裂纹信息，本文提出了两种分形的方法对路面破损图像中的裂纹信息进行识别，一种是分形的自相似性对路面破损图像的识别，另外一种是基于分形理论中离散分数布朗运动模型进行识别。研究了如何把破损图像中的标志标线和车辙等信息去除以及对破损路面图像中裂纹信息的增强与图像分割，最终实现了路面破损图像中的裂纹识别的研究目的。本论文主要研究成果建立的分形模型能很好的描述具有纹理的图像；对采集的路面图像进行了预处理以达到改善图像质量的目的，为最后路面破损图像裂纹的边缘检测与自动识别做准备；针对路面破损图像的自相似和自仿射性的特点，，根据分形理论对路面破损图像中裂纹进行边缘识别检测。根据路面破损图像的分形特性，提出了基于分形的自相似性和DFBR模型对路面破损图像的裂纹识别，从而识别出路面破损图像的裂纹信息。
[Abstract]:At present, the maintenance and management of roads both at home and abroad are carried out mainly according to the data and information from manual detection and automatic testing. Therefore, great costs are put into the research and development of various road automatic acquisition equipment and data processing systems at home and abroad. The smoothness and environmental information around the road, followed by analysis and recognition of the collected data, so as to provide maintenance and maintenance reports. The information processing of pavement damage data processing, due to the complexity of the road surface itself and the influence of the external environment of the data during the collection process, makes the data processing of the damaged image of the pavement. The workload is huge and takes a long time. Therefore, it is of great significance to study the automatic image recognition system for pavement distress.
The content of this study is to automatically recognize the damaged image of the pavement which is collected by the equipment. It is mainly to recognize all kinds of cracks in the damaged image. In order to identify the crack information in the damaged image, this paper proposes two fractal methods to identify the crack information in the damaged image of the pavement. The fractal self similarity is used to recognize the damaged image of the pavement, and the other is based on the discrete fractional Brown motion model in fractal theory. The information of marking and rutting in the damaged image and the enhancement of the crack information in the damaged road image and the image segmentation are studied, and the pavement damage is finally realized. The purpose of research on crack identification in images.
The fractal model of the main research results in this paper can describe the image with texture well, preprocessing the image of the collected pavement to improve the image quality, and prepare the edge detection and automatic recognition for the crack of the broken road surface, and the self similarity and self affine of the damaged road surface. On the basis of fractal theory, the fractal theory is used to detect the cracks in the damaged image of the pavement. According to the fractal characteristics of the damaged image of the pavement, a fractal based self similarity and DFBR model is proposed to identify the cracks in the damaged image of the pavement, thus identifying the crack information of the damaged image of the pavement.
【学位授予单位】：长安大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U495;U416.0

【参考文献】