Toward Accurate Cultural Asset Digitization: Analyzing on Object-centric 3D Reconstruction

High-fidelity 3D reconstruction is essential for the preservation, restoration, and digital archiving of cultural assets. However, state-of-the-art methods such as Neural Radiance Fields (NeRFs) and 3D Gaussian Splatting (3DGS) are sensitive to background elements, which can degrade the quality of the results. This challenge is particularly pronounced for cultural assets with intricate patterns and material properties, where inaccurate background segmentation can lead to significant artifacts. This study aims to systematically analyze the impact of background segmentation precision on the quality of 3D reconstruction. To this end, we develop an Iterative Segmentation technique to generate masks with varying accuracies by progressive refinement. By applying these intermediate masks to the 3DGS reconstruction at each step, we quantitatively demonstrate the overall improvement in the reconstruction quality as the segmentation accuracy increases. Furthermore, using the final refined mask as a baseline, we conduct a controlled study to compare the distinct effects of intentional “under-segmentation” (background leakage) and “over-segmentation” (object erosion) errors on the geometric and photometric fidelities of the 3D model. This study provides a quantitative analysis of the effect of different segmentation error types on the reconstruction of cultural assets, underscoring the critical need for precise foreground isolation to achieve high-fidelity results.