基于图像识别处理的桥梁底面裂缝检测方法的研究

发布时间：2018-05-28 14:49

本文选题：桥梁安全 + 裂缝识别　；参考：《北京交通大学》2016年硕士论文

【摘要】：高速列车是当今客运专线上的主要工具。桥梁作为高速铁路的主要载体,其安全性对于行车运输十分重要。桥梁裂缝检测是桥梁检测中的重要部分。目前桥梁裂缝的检测主要依赖于人工检测,这种方式费时费力,且危险性较高。基于图像识别处理的桥梁裂缝检测技术改善了这些不足,并已然成为一个非常活跃的研究课题。本文通过数字图像处理技术,提出了裂缝图像中裂缝识别和裂缝测量的改进算法。在裂缝识别方面,主要工作和创新点如下所述：(1)通过投影法对裂缝图像进行水平垂直投影,分析投影图像特征,初步确定图像中包含裂缝信息的目标区域；(2)对初步确定的目标区域进行区域生长,生长后的区域图像只包含裂缝信息和背景信息,从而达到裂缝识别；(3)投影法与区域生长相结合,改善了传统裂缝图像识别中过程较为繁琐的不足,直接从裂缝区域出发进行图像分割,识别速度较快,准确度较高,适用于现场实际裂缝检测。在裂缝测量方面,主要工作和创新点如下：(1)通过投影法对横向、纵向、斜向裂缝进行水平垂直投影,统计投影图像中像素点个数,通过每个像素点的实际大小计算裂缝宽度,长度；(2)对网状裂缝进行投影,并计算网状裂缝面积及空间占用率；(3)根据本算法对实际裂缝图像进行分类测量并统计数据,与传统的人工测量进行比较,分析算法的优劣性。本文改进了基于图像处理的裂缝检测技术中裂缝识别和裂缝测量的算法。通过实验结果分析,该方法在速度及准确性上均具有一定的优势,从而更有效地对桥梁底面进行裂缝检测。
[Abstract]:High-speed train is the main tool of passenger dedicated line. As the main carrier of high-speed railway, the safety of bridge is very important for transportation. Bridge crack detection is an important part of bridge detection. At present, bridge crack detection mainly depends on manual detection, which is time-consuming and dangerous. The bridge crack detection technology based on image recognition improves these shortcomings and has become a very active research topic. In this paper, an improved algorithm of crack recognition and crack measurement in crack image is proposed by digital image processing technology. In the area of crack identification, the main work and innovations are as follows: 1) projection method is used to project the crack image horizontally and vertically, and the characteristics of the projection image are analyzed. The target area that contains crack information in the preliminary determination image is used to grow the area of the initial target area, and the growing area image only contains crack information and background information. Thus, the projection method of crack recognition can be combined with the growth of the region, and the process of traditional crack image recognition is improved. The image segmentation is carried out directly from the crack area, and the recognition speed is faster and the accuracy is higher. It is suitable for actual crack detection on the spot. In the area of crack measurement, the main work and innovations are as follows: 1) horizontal and vertical projection of transverse, longitudinal and oblique cracks through projection method, statistics of the number of pixels in the projection image, and calculation of crack width through the actual size of each pixel point. Based on this algorithm, the actual crack images are classified and measured, and compared with the traditional manual measurement, the advantages and disadvantages of the algorithm are analyzed. In this paper, the algorithm of crack identification and crack measurement based on image processing is improved. The experimental results show that the method has some advantages in speed and accuracy, so it is more effective to detect the cracks at the bottom of the bridge.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：U446.3;TP391.41

【相似文献】