• Smart Structures and Systems
• /
• 제33권2호
• /
• pp.105-118
• /
• 2024

In order to realize tiny bridge crack discovery by UAV-based machine vision, a novel method combining deep learning and tensor voting is proposed. Firstly, the grid images of crack are detected and descripted based on SE-ResNet50 to generate feature points. Then, the probability significance map of crack image is calculated by tensor voting with feature points, which can define the direction and region of crack. Further, the crack detection anchor box is formed by non-maximum suppression from the probability significance map, which can improve the robustness of tiny crack detection. Finally, a case study is carried out to demonstrate the effectiveness of the proposed method in the Xiangjiang-River bridge inspection. Compared with the original tensor voting algorithm, the proposed method has higher accuracy in the situation of only 1-2 pixels width crack and the existence of edge blur, crack discontinuity, which is suitable for UAV-based bridge crack discovery.

• 보존과학회지
• /
• 제27권3호
• /
• pp.251-260
• /
• 2011

이 연구에서는 석조문화유산의 손상유형별 표준범례를 제시하고, 손상지도 작성방법에 대한 공정시스템을 구축하였으며, 균열지수 개발과 표면 및 3차원 손상율 평가기법을 제시하였다. 손상유형별 표준범례는 균열, 박리, 박락, 탈락, 입상분해 및 공동으로 세분한 다음 상용 그래픽 프로그램으로 제작하였으며, 손상지도는 손상 영역에 대한 정확도와 신뢰도를 높이기 위해 3차원 디지털복원과 고해상도 사진맵핑 기술을 적용하였다. 또한 균열지수를 개발하여 대상 석조문화유산의 물리적 손상도에 대한 정량평가를 수행하였고, 가상복원 모델링을 통해 탈락부의 부피와 3차원 손상율을 산출하였다. 이를 통해 마곡사오층석탑의 손상도를 정량적으로 평가한 결과, 전체적으로 북측면이 구조상 균열(1.70), 미세균열(1.34), 박락(20.2%), 탈락(13.0%)의 손상점유율이 높게 나타났으며, 1층 옥개석의 3차원 손상율은 6.7%로 산출되었다.

• 산업기술연구
• /
• 제28권A호
• /
• pp.89-96
• /
• 2008

An increase in the amount of cracking in repaired concrete bridge decks using VES-LMC(Very Early Strength - Latex Modified Concrete ; below VES-LMC) has been noticed by Yun et al(1). Literature indicates that indeed many concrete bridge decks develop transverse cracking, most developing at early ages(3~7 days), many right after construction. The purpose of this study was to establish prevention of map, transverse and longitudinal cracking in VES-LMC and to provide a control methods for minimizing the occurrence of cracks. The proposed prevention against map and transverse cracking was verified by field applications. VES cement was modified, the unit cement contents was reduced into $360kg/m^3$ from $390kg/m^3$, the maximum size of coarse aggregate was increase into 19mm from 13mm, wire mesh and steel fibers were incorporated in concrete mixture. A series of variable combinations were attempted. As a results, the proposed prevention against map and transverse cracking was verified because no crack were occurred until 90 days after overlay.

• 한국구조물진단유지관리공학회 논문집
• /
• 제13권3호통권55호
• /
• pp.173-180
• /
• 2009

초속경 라텍스개질 콘크리트는 교량바닥판 보수 후 조기교통개방을 가능하도록 하기위해 개발되었다. 본 논문의 목적은 초속경 라텍스개질 콘크리트에 발생하는 망상형, 횡방향 및 종방향 균열에 대한 원인을 분석하여 균열발생을 최소화 할 수 있는 방안을 마련하고, 현장 시험시공을 통하여 균열 억제방안을 검증하는 것이다. 횡방향 균열발생을 최소화하기 위하여 시멘트 성능의 개선과 더불어 단위시멘트량을 390kg/$m^3$에서 360kg/$m^3$으로 줄이고 굵은 골재의 최대치수를 13mm에서 19mm로 변경하였다. 시공측면에서 망상형 균열발생을 억제하기 위하여 강섬유와 와이어 메시를 사용하였고, 콘크리트 타설 직후 양생이 이뤄질 수 있도록 하였다. 검증실험 대상교량의 현장 균열조사결과 미세한 크기의 횡방향 균열과 종방향 균열을 제외하면, 3년 동안 구조적 균열이 발생하지 않은 것으로 조사되었다. 따라서 제안된 균열억제 방안이 균열억제에 효과적임을 확인하였다.

• 한국공작기계학회논문집
• /
• 제12권2호
• /
• pp.65-70
• /
• 2003

In this study, researchers developed the estimative algorithm for artificial defect in semiconductor packages and performed it by pattern recognition technology. For this purpose, the estimative algorithm was included that researchers made software with MATLAB. The software consists of some procedures including ultrasonic image acquisition, equalization filtering, Self-Organizing Map and Backpropagation Neural Network. Self-organizing Map and Backpropagation Neural Network are belong to methods of Neural Networks. And the pattern recognition technology has applied to classify three kinds of detective patterns in semiconductor packages : Crack, Delamination and Normal. According to the results, we were confirmed that estimative algerian was provided the recognition rates of 75.7% (for Crack) and 83.4% (for Delamination) and 87.2 % (for Normal).

• 대한건축학회논문집:구조계
• /
• 제36권3호
• /
• pp.113-120
• /
• 2020

The purpose of this study is to compare the models of Deep Learning-based Convolution Neural Network(CNN) for concrete crack detection. The comparison models are AlexNet, GoogLeNet, VGG16, VGG19, ResNet-18, ResNet-50, ResNet-101, and SqueezeNet which won ImageNet Large Scale Visual Recognition Challenge(ILSVRC). To train, validate and test these models, we constructed 3000 training data and 12000 validation data with 256×256 pixel resolution consisting of cracked and non-cracked images, and constructed 5 test data with 4160×3120 pixel resolution consisting of concrete images with crack. In order to increase the efficiency of the training, transfer learning was performed by taking the weight from the pre-trained network supported by MATLAB. From the trained network, the validation data is classified into crack image and non-crack image, yielding True Positive (TP), True Negative (TN), False Positive (FP), False Negative (FN), and 6 performance indicators, False Negative Rate (FNR), False Positive Rate (FPR), Error Rate, Recall, Precision, Accuracy were calculated. The test image was scanned twice with a sliding window of 256×256 pixel resolution to classify the cracks, resulting in a crack map. From the comparison of the performance indicators and the crack map, it was concluded that VGG16 and VGG19 were the most suitable for detecting concrete cracks.

• 한국공작기계학회:학술대회논문집
• /
• 한국공작기계학회 2005년도 춘계학술대회 논문집
• /
• pp.91-96
• /
• 2005

In this study, researchers developed the estimative algorithm for artificial defects in semiconductor packages and performed it by pattern recognition technology. For this purpose, the estimative algorithm was included that researchers made software with MATLAB. The software consists of some procedures including ultrasonic image acquisition, equalization filtering, Self-Organizing Map and Backpropagation Neural Network. Self-Organizing Map and Backpropagation Neural Network are belong to methods of Neural Networks. And the pattern recognition technology has applied to classify three kinds of detective patterns in semiconductor packages: Crack, Delamination and Normal. According to the results, we were confirmed that estimative algorithm was provided the recognition rates of $75.7\%$ (for Crack) and $83_4\%$ (for Delamination) and $87.2\%$ (for Normal).

• Structural Monitoring and Maintenance
• /
• 제1권3호
• /
• pp.309-322
• /
• 2014

Detection of fatigue cracks at an early stage of their development is important in structural health monitoring. The breathing of cracks in a structure generates higher harmonic components of the exciting frequency in the frequency spectrum. Previously, the residual operational deflection shape (R-ODS) method was successfully applied to beams with a single crack. The method is based on the ODSs at the exciting frequency and its higher harmonic components which consider both amplitude and phase information of responses to map the deflection pattern of structures. Although the R-ODS method shows the location of a single crack clearly, its identification for the location of multiple cracks in a structure is not always obvious. Therefore, an improvement to the R-ODS method is presented here to make the identification process distinct for the beams with multiple cracks. Numerical and experimental examples are utilised to investigate the effectiveness of the improved method.

• Smart Structures and Systems
• /
• 제32권3호
• /
• pp.135-151
• /
• 2023

Using 3D asphalt pavement surface data, a deep and multiscale network named CrackNet-M is proposed in this paper for pixel-level crack detection for improvements in both accuracy and robustness. The CrackNet-M consists of four function-specific architectural modules: a central branch net (CBN), a crack map enhancement (CME) module, three pooling feature pyramids (PFP), and an output layer. The CBN maintains crack boundaries using no pooling reductions throughout all convolutional layers. The CME applies a pooling layer to enhance potential thin cracks for better continuity, consuming no data loss and attenuation when working jointly with CBN. The PFP modules implement direct down-sampling and pyramidal up-sampling with multiscale contexts specifically for the detection of thick cracks and exclusion of non-crack patterns. Finally, the output layer is optimized with a skip layer supervision technique proposed to further improve the network performance. Compared with traditional supervisions, the skip layer supervision brings about not only significant performance gains with respect to both accuracy and robustness but a faster convergence rate. CrackNet-M was trained on a total of 2,500 pixel-wise annotated 3D pavement images and finely scaled with another 200 images with full considerations on accuracy and efficiency. CrackNet-M can potentially achieve crack detection in real-time with a processing speed of 40 ms/image. The experimental results on 500 testing images demonstrate that CrackNet-M can effectively detect both thick and thin cracks from various pavement surfaces with a high level of Precision (94.28%), Recall (93.89%), and F-measure (94.04%). In addition, the proposed CrackNet-M compares favorably to other well-developed networks with respect to the detection of thin cracks as well as the removal of shoulder drop-offs.

• 대한기계학회논문집A
• /
• 제25권12호
• /
• pp.2030-2039
• /
• 2001

Recently, various inpipe inspection robots are developed and its effective values are increased in industrial use. However, it is so difficult to make a inpipe inspection robot system which has flexible mobility and accuracy of inspection in pipelines. Especially, it is very important to know the exact crack position. In this paper, we are to present a lately developed inpipe inspection robot system which can resolve the above Problems. The robot is configured as an articulated structure like a snake. Two active driving vehicles are located in front and rear of the inspection robot respectively and passive modules such as a nondestructive testing module and a control module are chained between the active vehicles. Special feature of the robot system is a ground interface, which is able to show informations of robot and pipelines. By using this, so called virtual map in this paper, user is able to know the pipelines'feature and crack position.