Off-campus UNL users: To download campus access dissertations, please use the following link to log into our proxy server with your NU ID and password. When you are done browsing please remember to return to this page and log out.
Non-UNL users: Please talk to your librarian about requesting this dissertation through interlibrary loan.
Nondestructive Evaluation of Sub-Surface Defect in Concrete Structures Based on High-Resolution Aerial Infrared Thermography
Infrared thermography (IRT), as one of the nondestructive-detection (NDT) methods, has been explored for decades for condition assessments of concrete structures. Mixed results were reported in detecting sub-surface defects, such as delamination for the bridge decks (Omar, Nehdi, & Zayed, 2018). While its value has been well-recognized, it is also widely acknowledged that this technology’s accuracy and robustness are highly susceptible to a wide range of environmental noises, which include weather conditions, and deck surfaces’ color, texture, roughness, moisture, etc. Conventional methods using the temperature contrast as the detection criterion performed poorly in the presence of unfavorable environmental noises. Additionally, conventional IRT methods highly rely on experts’ manual interpretation of the IRT images, which is time-consuming, inconsistent, and unreliable in terms of the interpretation of the result. Tapping recent developments in computer vision, image processing, deep learning, and high-resolution aerial IRT, spatial aerial IRT features of the concrete deck’s defects were investigated experimentally and numerically under both lab and field settings. Novel IRT segmentation approaches of underlying concrete defects are developed and tested to enhance the accuracy, robustness of IRT, and to enable the automation of detection. In particular, several image processing techniques are investigated to understand the defect's characteristics, such as the blob-like feature of delamination (chapter 3), the morphological definition for defect extraction (chapter 4), and temperature gradient features on the defect boundary for defect segmentation (chapter 4). To achieve the automation of result inferencing, a deep learning model based on encoder-decoder architecture is proposed (chapter 5). The results from both lab and field experiments demonstrated that the developed novel image processing methods using aerial IRT’s spatial characteristic of the defect significantly outperform the state-of-the-art approaches in the detection and segmentation performances when handling the non-uniform background in the thermal images in terms of accuracy, robustness, and automation.
Engineering|Materials science|Architectural engineering
Cheng, Chongsheng, "Nondestructive Evaluation of Sub-Surface Defect in Concrete Structures Based on High-Resolution Aerial Infrared Thermography" (2020). ETD collection for University of Nebraska - Lincoln. AAI28085882.