Generating of damage maps of building façades through digital image processing captured by drones and use of digital photogrammetry

Abstract The aim of this study was to apply a methodology for automatic generation of maps of damages in orthoimages of façades, using UAVs to capture the photographs. The generation of maps of damages aimed to assist in the visual inspection method used in activities of maintenance, restoration or intervention in buildings. Using the dense stereo matching (DSM) method, geometric models of buildings weren generated by the point cloud in the PhotoScan program for obtaining orthoimages. After this step, the orthoimages were submitted to generation of maps of damages of staining in Tyche (2015) software, which mapped and quantified in percentage terms these defects. The results of the maps of damage were also classified qualitatively. The application of UAV proved to be an efficient tool for façade inspection activity and for capturing images to be used by DSM. The generation of maps of damages by Tyche (2015) proved also effective, however it was verified that the user interaction was important to achieve the expected results.摘要 本研究旨在应用一套方法，基于无人机（Unmanned Aerial Vehicle，UAV）采集的建筑立面正射影像（orthoimage）实现损伤地图的自动生成，以此辅助建筑维护、修复或改造活动中的目视检测工作。本研究采用密集立体匹配（dense stereo matching，DSM）方法，通过PhotoScan软件基于点云构建建筑几何模型以获取正射影像。随后，将所得正射影像导入Tyche (2015)软件，生成污渍类损伤地图，该软件可对这类缺陷进行标注并以百分比形式量化其占比。同时，对损伤地图的生成结果开展定性分类。实践证明，无人机可作为高效工具用于建筑立面检测及为密集立体匹配方法采集图像。Tyche (2015)软件的损伤地图生成方案同样成效显著，但研究证实，为达成预期结果，用户的交互操作不可或缺。