Seite - 779 - in Book of Full Papers - Symposium Hydro Engineering

3. SURFACE MONITORING SOLUTION WITH TOTAL STATIONS FOR ENHANCEMENT OF VISUAL INSPECTIONS 3.1. METHODOLOGY Large dams commonly show concrete deficiencies, which do not necessarily pose a risk to the structure’s safety. Some of these deficiencies appear just after construction and do not change over time. Therefore, detection of any visual changes on the dam’s surface is crucial. The objective is the development of an algorithm to automatically detect new defects and identify the extent of deterioration of existing ones. In principle, we exploit the spatial information of the TS and use images to derive semantic information for visual change detection. We divide our evaluation into two parts: the on-site detection and acquisition and the change detection analysis, which we perform in a subsequent step (see Fig. 6). We developed an application to carry out all required work processes (remote control, processing) on-site on a standard laptop. After the user measures the contours of the dam’s surface, the instrument starts full dam photo acquisition with the OVC. With image-processing techniques applied to the OVC images, we localize defect regions on the dam (cf. Section 3.2). The known camera model simplifies the combination of image content to spatial information, i.e. horizontal (Hz) and vertical (V) angles. Fig 6: Operating principle of the proposed surface monitoring solution Afterwards, the application starts high-resolution scan data and OAC photo acquisition of potential defect regions. At this point, the operator can manually interact in order to minimize the time effort needed for data acquisition. However, to enhance the entire acquisition process, we propose an automatic classification procedure based on foreknowledge. Therefore, we establish a database containing coordinates of true defects as well as mistakenly identified defect regions in the 779