This paper proposes an automated crack evaluation technique for a high-rise bridge pier using a climbing robot. The proposed technique enables to automatically detect and quantify the bridge pier cracks even where cannot easily access by human for visual inspection. To achieve it, high quality images are obtained by scanning the vision cameras embedded on the climbing robot along the bridge pier surface. Then, a feature extraction-based image stitching algorithm is newly developed and applied for establishing the entire region of interest (ROI) images. The ROI images are then processed with a semantic segmentation algorithm for automated crack detection. Finally, the detected cracks are precisely quantified by a crack quantification algorithm. The proposed technique is validated using in-situ test data obtained from Jang-Duck bridge located at Gangneung city, South Korea. The test results reveal that the proposed technique successfully evaluate the bridge pier cracks with precision of 90.92 % and recall of 97.47 %.

This paper proposes a deep learning-based underground object classification technique incorporated with phase analysis of ground penetrating radar (GPR) for enhancing the underground object classification capability. Deep convolutional neural network (CNN) using the combination of the B- and C-scan images has recently emerged for automated underground object classification. However, it often leads to misclassification because arbitrary underground objects may have similar signal features. To overcome the drawback, the combination of B- and C-scan images as well as phase information of GPR are simultaneously used for CNN in this study, enabling to have more distinguishable signal features among various underground objects. The proposed technique is validated using in-situ GPR data obtained from urban roads in Seoul, South Korea. The validation results show that the false alarm is significantly reduced compared to the CNN results using only B- and C-scan images.

This paper presents an automated determination technique of optimal subset sizes for digital image correlation (DIC) analysis of speckle patterned images. The smaller subset size would typically have the higher DIC accuracy with respect to local minute deformation, but insufficient speckle pattern information within the excessively small often augment DIC errors due to lack of correlation features. Therefore, optimal subset size determination is crucial for the precise DIC analysis. To automate the optimal subset size determination process, first, the reference and test speckle pattern images are obtained from the target structure surface with a certain time interval. Then, an initial seed point which will be used for the subset center point is assigned on the reference speckle pattern image. Subsequently, normalized cross correlation (NCC) between the reference and test images is performed by increasing subset sizes from the seed point. Next, the matching distance between the two images is calculated using the maximum correlation coefficient. As the subset size increases, the matching distance between the two subsets converges a certain value. It physically means that the sufficient correlation features will be included in the subset. Finally, the optimal subset size can be determined by selecting the minimum subset size where the matching distance value starts to be converged. The proposed technique is experimentally validated using an aluminium plate with sprayed speckle pattern.

Fiber-Reinforced Cementitious Composites (FRCCs)는 시멘트 복합체에 혼입한 전도성 섬유로 인해 전기 전도성을 가진다. 이러한 특성은 전기적 응답 계측을 통하여 별도의 센서 설치가 필요 없는 구조물의 균열 모니터링을 가능하게 한다. 하지만 전기적 응답은 균열 발생뿐 만 아니라 온도의 변화에도 민감하게 변화하기 때문에 온도 요인은 전기적 응답 계측을 통한 균열 탐지를 방해하는 요소로 작용할 수 있다. 더욱이 전기적 응답을 측정하기 위한 탐침의 개수가 증가 할수록 원하지 않은 접촉 노이즈가 발생하기 때문에 이 논문에서는 탐침의 개수를 줄이기 위해 자체적인 자가센싱 임피던스 회로를 설계하였다. FRCC의 균열 발생과 온도 변화가 임피던스에 미치는 영향성은 자가센싱 임피던스 회로를 이용해 실험적으로 측정되었으며, 실험 결과, 임피던스 응답은 균열 발생보다 온도 변화에 더 민감하게 변화됨을 알 수 있었다.

강섬유 보강 자기충전 콘크리트(Steel Fiber Reinforced Self-Compacting Concrete, SFRSCC)는 사회기반 시설이나 초고층 빌딩, 원자력 발전 시설, 병원, 댐, 수로 등 전반적으로 널리 사용되어지고 있는 재료이다. SFRSCC는 짧고, 개별적인 보강 섬유로 인해 일반적인 자기충전 콘크리트(Self-Compacting Concrete, SCC) 보다 인장 강도, 연성, 휨 강성 등에서 뛰어난 성능을 보인다. 하지만 SFRSCC의 이러한 성능은 섬유의 방향성에 의해 크게 좌우되는 경향이 있다. 짧고 개별적인 섬유들은 타설 과정에서 섬유의 방향성을 컨트롤 할 수 없기 때문에 무분별하게 콘크리트 내에 위치하게 된다. 섬유의 방향이 제어되지 않은 상태에서 콘크리트의 경화가 진행될 경우 휨 강성과 인장 강도의 저하를 야기하고, 이는 예상 강도 미달의 원인이 될 수 있기 때문에 SFRSCC를 사용할 때 섬유의 정렬은 중요한 요소가 된다. 따라서 본 연구에서는 유한 요소법을 사용하여 타설 공정 중 콘크리트 매트리스의 점도 및 입구 속도가 섬유 방향에 미치는 영향에 대해 분석하였다.

This paper proposes a deep learning-based crack evaluation technique using hybrid images. The use of the hybrid images combining vision and infrared images are able to improve crack detectability while minimizing false alarms. In particular, large-scale infrastructures can be inspected by an UAV-mounted hybrid image scanning (HIS)　system, and the corresponding huge amount of data is typically difficult to be analyzed by experts. To automate such making-decision process, deep convolutional neural network is used in this study. As the very first stage, a lab-scale HIS system is developed using a scanning zig and experimentally validated using a concrete specimen with various-size cracks. The test results reveal that macro- and micro-cracks are successfully and automatically detected with minimizing false-alarms.