ScholarWorks@Y-Scholar Hub

초록

This study developed a web-based tool for effectively generating and managing spatial visual information from panoramic images of interior spaces. Panoramic images serve as comprehen- sive visual data that capture spaces in all directions, playing a vital role in architectural docu- mentation and interior design visualization. However, existing viewers and tools demonstrate limitations in generating diverse visual information when converting equirectangular projection formats into naturalistic viewpoints. The proposed tool implements spherical coordinate transfor- mation algorithms and artificial intelligence-based language models to automatically generate multi-viewpoint images and sequential footage, while simultaneously providing spatial descrip- tions for the corresponding visual data. Users can customize and manage visual assets by manipulating parameters including field of view, horizontal and vertical rotation angles, and res- olution settings. The generated data can be integrated into spatial databases for extended appli- cations. While primarily designed for interior spaces, this tool's functionality extends to exterior panoramic imagery, presenting novel approaches for the generation and management of spatial visual information.