Path design for ground-based panoramic image acquisition

Recently, panoramic images have been used for map services, urban planning, and disaster monitoring. Moreover, there are many studies related with 3D modeling, augmented reality application, building information modeling, and construction information modeling. In particular, a 3D model generation using from images is widely applied in a mapping, structure monitoring, and cultural heritage surveys. In conventional approaches, we acquired aerial images with constant overlaps for 3D modeling. Moreover, images are acquired from various viewpoints to reconstruct a 3D model without occlusion to avoid occlusions. However, a measurement of complex objects requires a huge amount of time for an image acquisition and processing. Therefore, in this research, we aimed to develop a methodology of path design for complex civil engineering structures, such as dams and bridges. First, we focused on a path design for 3D modeling using panoramic images. A panoramic image could be assumed as integrated images taken by some cameras. Thus, we prepared images taken by some cameras for our experiments. Second, we evaluated some path designs based on Structure from Motion in our experiments. Through our experiments, we confirmed that the efficiency in a panoramic image acquisition depended on a camera path design.