Pixel-Wise Crack Detection Using Deep Local Pattern Predictor for Robot Application
Robotic vision-based crack detection in concrete bridges is an essential task to preserve these assets and their safety. The conventional human visual inspection method is time consuming and cost inefficient. In this paper, we propose a robust algorithm to detect cracks in a pixel-wise manner from r...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2018-09-01
|
Series: | Sensors |
Subjects: | |
Online Access: | http://www.mdpi.com/1424-8220/18/9/3042 |
id |
doaj-261a8ea8af304f378e498c521da6d880 |
---|---|
record_format |
Article |
spelling |
doaj-261a8ea8af304f378e498c521da6d8802020-11-24T21:21:10ZengMDPI AGSensors1424-82202018-09-01189304210.3390/s18093042s18093042Pixel-Wise Crack Detection Using Deep Local Pattern Predictor for Robot ApplicationYundong Li0Hongguang Li1Hongren Wang2School of Electronic and Information Engineering, North China University of Technology, Beijing 100144, ChinaUnmanned Systems Research Institute, Beihang University, Beijing 100191, ChinaSchool of Electronic and Information Engineering, Beihang University, Beijing 100191, ChinaRobotic vision-based crack detection in concrete bridges is an essential task to preserve these assets and their safety. The conventional human visual inspection method is time consuming and cost inefficient. In this paper, we propose a robust algorithm to detect cracks in a pixel-wise manner from real concrete surface images. In practice, crack detection remains challenging in the following aspects: (1) detection performance is disturbed by noises and clutters of environment; and (2) the requirement of high pixel-wise accuracy is difficult to obtain. To address these limitations, three steps are considered in the proposed scheme. First, a local pattern predictor (LPP) is constructed using convolutional neural networks (CNN), which can extract discriminative features of images. Second, each pixel is efficiently classified into crack categories or non-crack categories by LPP, using as context a patch centered on the pixel. Lastly, the output of CNN—i.e., confidence map—is post-processed to obtain the crack areas. We evaluate the proposed algorithm on samples captured from several concrete bridges. The experimental results demonstrate the good performance of the proposed method.http://www.mdpi.com/1424-8220/18/9/3042local pattern predictorcrack detectionbridge inspectionconvolutional neural networksrobotic vision |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Yundong Li Hongguang Li Hongren Wang |
spellingShingle |
Yundong Li Hongguang Li Hongren Wang Pixel-Wise Crack Detection Using Deep Local Pattern Predictor for Robot Application Sensors local pattern predictor crack detection bridge inspection convolutional neural networks robotic vision |
author_facet |
Yundong Li Hongguang Li Hongren Wang |
author_sort |
Yundong Li |
title |
Pixel-Wise Crack Detection Using Deep Local Pattern Predictor for Robot Application |
title_short |
Pixel-Wise Crack Detection Using Deep Local Pattern Predictor for Robot Application |
title_full |
Pixel-Wise Crack Detection Using Deep Local Pattern Predictor for Robot Application |
title_fullStr |
Pixel-Wise Crack Detection Using Deep Local Pattern Predictor for Robot Application |
title_full_unstemmed |
Pixel-Wise Crack Detection Using Deep Local Pattern Predictor for Robot Application |
title_sort |
pixel-wise crack detection using deep local pattern predictor for robot application |
publisher |
MDPI AG |
series |
Sensors |
issn |
1424-8220 |
publishDate |
2018-09-01 |
description |
Robotic vision-based crack detection in concrete bridges is an essential task to preserve these assets and their safety. The conventional human visual inspection method is time consuming and cost inefficient. In this paper, we propose a robust algorithm to detect cracks in a pixel-wise manner from real concrete surface images. In practice, crack detection remains challenging in the following aspects: (1) detection performance is disturbed by noises and clutters of environment; and (2) the requirement of high pixel-wise accuracy is difficult to obtain. To address these limitations, three steps are considered in the proposed scheme. First, a local pattern predictor (LPP) is constructed using convolutional neural networks (CNN), which can extract discriminative features of images. Second, each pixel is efficiently classified into crack categories or non-crack categories by LPP, using as context a patch centered on the pixel. Lastly, the output of CNN—i.e., confidence map—is post-processed to obtain the crack areas. We evaluate the proposed algorithm on samples captured from several concrete bridges. The experimental results demonstrate the good performance of the proposed method. |
topic |
local pattern predictor crack detection bridge inspection convolutional neural networks robotic vision |
url |
http://www.mdpi.com/1424-8220/18/9/3042 |
work_keys_str_mv |
AT yundongli pixelwisecrackdetectionusingdeeplocalpatternpredictorforrobotapplication AT hongguangli pixelwisecrackdetectionusingdeeplocalpatternpredictorforrobotapplication AT hongrenwang pixelwisecrackdetectionusingdeeplocalpatternpredictorforrobotapplication |
_version_ |
1726000590992965632 |