Research on multi-view 3D reconstruction algorithm of ancient buildings based on deep learning
As a treasure of human civilization, the digital protection of ancient buildings is of great significance. Aiming at the problems of texture occlusion and high computational complexity in the reconstruction of ancient buildings by traditional 3D reconstruction methods, this paper proposes a multi-vi...
| Опубликовано в: : | Инноватика-2025 : сборник материалов XXI Международной школы-конференции студентов, аспирантов и молодых ученых, 28-30 апреля 2025 г., г. Томск, Россия С. 533-537 |
|---|---|
| Главный автор: | |
| Формат: | Статья в сборнике |
| Язык: | English |
| Предметы: | |
| Online-ссылка: | https://vital.lib.tsu.ru/vital/access/manager/Repository/koha:001272937 Перейти в каталог НБ ТГУ |
| Итог: | As a treasure of human civilization, the digital protection of ancient buildings is of great significance. Aiming at the problems of texture occlusion and high computational complexity in the reconstruction of ancient buildings by traditional 3D reconstruction methods, this paper proposes a multi-view 3D reconstruction algorithm based on improved MPMNet. By constructing an ancient building dataset containing more than 8,000 images, combining the Delaunay triangulation algorithm to optimize the depth map initialization, and introducing a data enhancement module, the reconstruction accuracy is significantly improved. Experimental results show that MPMNet has higher integrity and detail restoration capabilities in the reconstruction of ancient buildings, while reducing memory consumption and processing time. |
|---|---|
| Библиография: | Библиогр.: 3 назв. |
| ISBN: | 9785936297311 |