3D Point Cloud Model Colorization by Dense Registration of Digital Images

A novel method for enhancing the photometric quality of 3D point cloud models has been developed, significantly improving colorization accuracy through dense registration of digital images. This approach addresses long-standing challenges in heritage documentation by treating image-to-geometry registration as a pose estimation problem, enabling automatic computation of both extrinsic and intrinsic camera parameters without prior camera information. The technique leverages photometric visual and virtual servoing (VVS) to project colors onto 3D models with higher resolution and fidelity. Validated through simulations and real-world experiments on Amiens Cathedral—both indoor and outdoor datasets—the method demonstrates substantial advancements in photometric quality, offering a robust solution for preserving and presenting architectural heritage with unprecedented detail.

Crombez, N., Caron, G., & Mouaddib, E. (2015). 3D POINT CLOUD MODEL COLORIZATION BY DENSE REGISTRATION OF DIGITAL IMAGES. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, XL-5/W4, 123–130. https://doi.org/10.5194/isprsarchives-xl-5-w4-123-2015