Depth-Consistent 3D Gaussian Splatting via Physical Defocus Modeling and Multi-View Geometric Supervision

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Deng, Yu, Zhao, Baozhu, Su, Junyan, Zhang, Xiaohan, Liu, Qi
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912706914680832
author Deng, Yu
Zhao, Baozhu
Su, Junyan
Zhang, Xiaohan
Liu, Qi
author_facet Deng, Yu
Zhao, Baozhu
Su, Junyan
Zhang, Xiaohan
Liu, Qi
contents Three-dimensional reconstruction in scenes with extreme depth variations remains challenging due to inconsistent supervisory signals between near-field and far-field regions. Existing methods fail to simultaneously address inaccurate depth estimation in distant areas and structural degradation in close-range regions. This paper proposes a novel computational framework that integrates depth-of-field supervision and multi-view consistency supervision to advance 3D Gaussian Splatting. Our approach comprises two core components: (1) Depth-of-field Supervision employs a scale-recovered monocular depth estimator (e.g., Metric3D) to generate depth priors, leverages defocus convolution to synthesize physically accurate defocused images, and enforces geometric consistency through a novel depth-of-field loss, thereby enhancing depth fidelity in both far-field and near-field regions; (2) Multi-View Consistency Supervision employing LoFTR-based semi-dense feature matching to minimize cross-view geometric errors and enforce depth consistency via least squares optimization of reliable matched points. By unifying defocus physics with multi-view geometric constraints, our method achieves superior depth fidelity, demonstrating a 0.8 dB PSNR improvement over the state-of-the-art method on the Waymo Open Dataset. This framework bridges physical imaging principles and learning-based depth regularization, offering a scalable solution for complex depth stratification in urban environments.
format Preprint
id arxiv_https___arxiv_org_abs_2511_10316
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Depth-Consistent 3D Gaussian Splatting via Physical Defocus Modeling and Multi-View Geometric Supervision
Deng, Yu
Zhao, Baozhu
Su, Junyan
Zhang, Xiaohan
Liu, Qi
Computer Vision and Pattern Recognition
Artificial Intelligence
Three-dimensional reconstruction in scenes with extreme depth variations remains challenging due to inconsistent supervisory signals between near-field and far-field regions. Existing methods fail to simultaneously address inaccurate depth estimation in distant areas and structural degradation in close-range regions. This paper proposes a novel computational framework that integrates depth-of-field supervision and multi-view consistency supervision to advance 3D Gaussian Splatting. Our approach comprises two core components: (1) Depth-of-field Supervision employs a scale-recovered monocular depth estimator (e.g., Metric3D) to generate depth priors, leverages defocus convolution to synthesize physically accurate defocused images, and enforces geometric consistency through a novel depth-of-field loss, thereby enhancing depth fidelity in both far-field and near-field regions; (2) Multi-View Consistency Supervision employing LoFTR-based semi-dense feature matching to minimize cross-view geometric errors and enforce depth consistency via least squares optimization of reliable matched points. By unifying defocus physics with multi-view geometric constraints, our method achieves superior depth fidelity, demonstrating a 0.8 dB PSNR improvement over the state-of-the-art method on the Waymo Open Dataset. This framework bridges physical imaging principles and learning-based depth regularization, offering a scalable solution for complex depth stratification in urban environments.
title Depth-Consistent 3D Gaussian Splatting via Physical Defocus Modeling and Multi-View Geometric Supervision
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2511.10316