Please use this identifier to cite or link to this item:
|Title:||Image-based Fine-scale Infrastructure Monitoring|
|Abstract:||Monitoring the physical health of civil infrastructure systems is an important task that must be performed frequently in order to ensure their serviceability and sustainability. Additionally, laboratory experiments where individual system components are tested on the fine-scale level provide essential information during the structural design process. This type of inspection, i.e., measurements of deflections and/or cracks, has traditionally been performed with instrumentation that requires access to, or contact with, the structural element being tested; performs deformation measurements in only one dimension or direction; and/or provides no permanent visual record. To avoid the downsides of such instrumentation, this dissertation proposes a remote sensing approach based on a photogrammetric system capable of three-dimensional reconstruction. The proposed system is low-cost, consists of off-the-shelf components, and is capable of reconstructing objects or surfaces with homogeneous texture. The scientific contributions of this research work address the drawbacks in currently existing literature. Methods for in-situ multi-camera system calibration and system stability analysis are proposed in addition to methods for deflection/displacement monitoring, and crack detection and characterization in three dimensions. The mathematical model for the system calibration is based on a single or multiple reference camera(s) and built-in relative orientation constraints where the interior orientation and the mounting parameters for all cameras are explicitly estimated. The methods for system stability analysis can be used to comprehensively check for the cumulative impact of any changes in the system parameters. They also provide a quantitative measure of this impact on the reconstruction process in terms of image space units. Deflection/displacement monitoring of dynamic surfaces in three dimensions is achieved with the system by performing an innovative sinusoidal fitting adjustment. The input data for the adjustment comes from either model-based image fitting or full surface fitting procedures. The crack characterization, i.e., estimation of the average crack width, approximate length and overall orientation, is achieved directly in three dimensions by detecting cracks in a region of interest in a truly-rectified photo via image processing techniques. This hybrid approach combines the use of both geometric and radiometric data, and it performs best in a multi-epoch setting.|
|Appears in Collections:||Electronic Theses|
Files in This Item:
|ucalgary_2016_detchev_ivan.pdf||Full thesis||28.38 MB||Adobe PDF||View/Open|
Items in The Vault are protected by copyright, with all rights reserved, unless otherwise indicated.