Crawling Method for Image-Based Space Matching in Digital Twin Smart Cities

To provide interactive services between the digital twin smart city and the actual smart city, space matching in physical space and virtual space is required. Visual place recognition, which is widely used for space matching, requires a 2D image database for 3D space because it identifies location information through matching between query images and database. We suggest a method of generating a database by collecting 2D images in 3D space for image-based spatial matching. In this paper, we introduce a crawling method that can collect continuous 2D images of the digital twin smart city. Our crawling method creates crawling units of a specific size and crawling points at regular intervals in the area where image crawling is required and generates the path for a virtual crawling device that can pass through all points on the unit. It makes a 2D image database that can express 3D space by crawling successive images obtained by rotating the virtual crawling device. By using our crawling method, we can help provide interactive services in the future smart city.