Investigation of Computer Vision Methods for the Building Construction Process.

摘要

The objective of this research was to investigate the feasibility of introducing computer vision methods to the building construction site. The primary use envisioned is direct input of data from a video to the computer for the purpose of productivity analysis. The results of processing actual footage from construction sites and other, existing buildings are described. There are observations made as to the effect of backlighting of the structure, lighting conditions and shadowing, and physical obstructions. Proposals are made as to methods to insure precise repeatable placement of observation cameras. The alternate proposal is translation of images obtained from different camera positions through the use of on-screen reference points. A summary of the physical barriers and technological problems, and suggested courses of action, is provided. Theses. (kt)