Monocular Online Reconstruction with Enhanced Detail Preservation

TL;DR

This paper introduces an online monocular 3D reconstruction framework that enhances detail preservation using Gaussian-based dense mapping, hierarchical management, and multi-level occupancy structures, achieving high-quality results efficiently.

Contribution

It presents a novel Gaussian-based dense mapping method with hierarchical management and multi-level occupancy structures for improved detail preservation in monocular online reconstruction.

Findings

Outperforms state-of-the-art RGB-only and RGB-D methods in reconstruction quality.

Maintains local and global consistency effectively.

Achieves high computational efficiency and scalability.

Abstract

We propose an online 3D Gaussian-based dense mapping framework for photorealistic details reconstruction from a monocular image stream. Our approach addresses two key challenges in monocular online reconstruction: distributing Gaussians without relying on depth maps and ensuring both local and global consistency in the reconstructed maps. To achieve this, we introduce two key modules: the Hierarchical Gaussian Management Module for effective Gaussian distribution and the Global Consistency Optimization Module for maintaining alignment and coherence at all scales. In addition, we present the Multi-level Occupancy Hash Voxels (MOHV), a structure that regularizes Gaussians for capturing details across multiple levels of granularity. MOHV ensures accurate reconstruction of both fine and coarse geometries and textures, preserving intricate details while maintaining overall structural integrity. Compared to state-of-the-art RGB-only and even RGB-D methods, our framework achieves superior reconstruction quality with high computational efficiency. Moreover, it integrates seamlessly with various tracking systems, ensuring generality and scalability.