Abstract:
Photogrammetry, a technique that enables the creation of 3D models from 2D images,
holds great potential for various applications. However, the current manual and time-
consuming nature of the process hinders its widespread adoption. This research aims
to address these challenges by developing a faster and more efficient photogrammetry
workflow. Traditionally, photogrammetry relied on stereo pairs of images to generate
digital elevation models. In this study, we propose a novel approach where multiple
images of an object captured from various angles are used, with the prerequisite that
the background scene remains static. These images are then loaded into our advanced
photogrammetry software. To expedite the 3D modelling process, our methodology
leverages mathematical algorithms and Al-driven learning-based techniques. By
combining the power of computational mathematics and artificial intelligence, we aim
to automate and optimize the reconstruction process, reducing the need for extensive
manual intervention. The resulting output will be a 3D model with lower density but
still possessing a high level of accuracy.
