Improving accuracy of automated 3-D building models for smart cities

Abstract

Photorealistic three-dimensional (3-D) models are important tools for urban and disaster management planning because they contain useful visual and spatial information for assessing the situation and responding accordingly. Thus, it is crucial for 3-D city models to maintain a high level of accuracy in portraying 3-D geometric objects. Despite that, significant research with respect to 3-D city models has been conducted; however, it is still difficult to create accurate 3-D representations, particularly across a wide area. Thus, this research is aimed at developing an automated 3-D city modeling application that utilizes a combination of aerial photographs, terrestrial light detection and ranging, and total station techniques for particularly dense urban areas. To enable the development of an automated 3-D city model, this research developed application software programmed in C++. This application enables users to generate 3-D images of buildings. Specifically, the 3-D city models are generated using the aerial photographs. The positional accuracy of the model is highly improved by comparing building models produced using a photogrammetric plotting instrument. Once created using a combination of aerial photos and terrestrial instruments, the 3-D city model is quantitatively assessed for completeness and suitability for commercial or public use.