A Quantitative Impact Analysis of Digital Image Resolution on Point Cloud Generation by Structure from Motion for Scaffolding Inspection

Abstract

Scaffolding is a temporary structure that directly affects the safety of a construction site. Frequent scaffolding inspection can raise working productivity and decrease the risk of casualties and losses. Current work procedure often forces the inspector to check the scaffolding closely, which is unsafety, time-consuming, and labor-intensive. Nowadays, numerous technologies can rapidly assist the inspection of the structure of interest such as Structure from Motion, which can create point clouds or models from a gigantic amount of digital images. However, although the quality of the output from Structure from Motion greatly relies on the quality of the input images, there is no clear data on the quantitative impact of input image quality on the output of point cloud generation. Therefore, this study is conducted to analyze the impact of image quality on the point cloud result, which is the fundamental input of the model assisting scaffolding inspection, and propose a guideline for site practitioners to be able to select appropriate digital image quality for Structure from Motion assisting scaffolding inspection.