[1]石帅,王睿,文思思,等.基于机器视觉的隧道衬砌裂缝图像分割处理算法研究[J].高速铁路技术,2020,11(01):17-22.[doi:10.12098/j.issn.1674-8247.2020.01.004]
　SHI Shuai,WANG Rui,WEN Sisi,et al.Research on Image Segmentation Algorithm of Tunnel Lining Based on Machine Vision[J].HIGH SPEED RAILWAY TECHNOLOGY,2020,11(01):17-22.[doi:10.12098/j.issn.1674-8247.2020.01.004]
点击复制

基于机器视觉的隧道衬砌裂缝图像分割处理算法研究()
分享到：

《高速铁路技术》[ISSN:1674-8247/CN:51-1730/U]

卷:
11卷
期数:
2020年01期
页码:
17-22
栏目:
出版日期:
2020-03-10

文章信息/Info

Title:
Research on Image Segmentation Algorithm of Tunnel Lining Based on Machine Vision
文章编号:
1674—8247(2020)01—0017-06
作者:
石帅;  王睿;  文思思;  汤盈盈
四川师范大学, 成都 610068
Author(s):
SHI Shuai;  WANG Rui;  WEN Sisi;  TANG Yingying
Sichuan Normal University, Chengdu 610068, China
关键词:
机器视觉|图像分割|Otsu法|最小误差法
Keywords:
machine vision|image segmentation|Otsu method|minimum error method
分类号:
TP391.91
DOI:
10.12098/j.issn.1674-8247.2020.01.004
文献标志码:
A
摘要:
基于机器视觉的裂缝检测方法因其高效率、低影响的优势已成为高速铁路隧道衬砌裂损检测的主要研究方向，但该法采集到的裂缝图像因隧道内部环境复杂、光照不均等原因导致图像质量很低，需经过图像分割（将裂缝区域与图像背景分离）才能有效提取裂缝的特征信息。图像分割处理中算法的选择与适用性非常关键，因此本文从Otsu法、迭代法、最小误差法、最大熵阈值分割法等传统分割算法的原理和运算步骤入手，借助MATLAB软件对采集的隧道衬砌裂缝图像进行模拟处理，得到各算法的处理图像。通过对各算法图像处理结果的对比分析可知：（1）4种算法都能较好地处理裂缝与背景分离明显的理想图像；（2）迭代法在处理背景颜色分布不均的图像时表现出优势；（3）最小误差法在处理与背景颜色相近的细微裂缝和灰度直方图接近正态分布概率密度曲线的图像占优势；（4）最大熵阈值分割法易受背景噪点干扰，总体处理结果不佳。
Abstract:
The crack detection method based on machine vision has become the main research direction of high-speed rail tunnel lining crack detection because of its high efficiency and low influence. However, the quality of crack image collected by this method is very low due to the complex internal environment of the tunnel and uneven illumination. Therefore, it is necessary to separate the crack area from the image background to effectively extract the feature information of the crack. The selection and applicability of the algorithm in image segmentation processing are very important. In this paper, the principles and operation steps of traditional segmentation algorithms, such as Otsu method, iterative method, minimum error method and maximum entropy method, are started, and with the help of MATLAB soft, the collected crack images of tunnel lining are simulated and processed to obtain the processing images of each algorithm. Through the comparative analysis of the image processing results of each algorithm, it can be seen that:(1) the four algorithms can better deal with the ideal images with obvious separation between crack and background; (2) the iterative method has advantages in dealing with the images with uneven background color distribution; (3) The minimum error method is dominant in dealing with the images with fine cracks similar to the background color and gray histogram close to the normal distribution probability density curve; (4) the maximum entropy method is easily disturbed by the background noise, so the overall processing results are not good.

备注/Memo

备注/Memo:
作者简介:石帅(2000-),男,在读本科生。基金项目:四川省大学生创新创业训练项目（项目编号：201810636328）引文格式:石帅, 王睿, 文思思, 等. 基于机器视觉的隧道衬砌裂缝图像分割处理算法研究[J]. 高速铁路技术,2020,11(1):17-22.SHI Shuai, WANGRui, WEN Sisi, et al. Research on Image Segmentation Algorithm of Tunnel Lining Based on Machine Vision[J]. High Speed Railway Technology, 2020, 11(1):17-22.
更新日期/Last Update: 2020-03-10