Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Dual-Level Precision Edges Guided Multi-View Stereo with Accurate Planarization

Kehua Chen, Zhenlong Yuan, Tianlu Mao, Zhaoqi Wang

TL;DR

The paper tackles robust 3D reconstruction in low-textured regions for multi-view stereo by addressing plane-model robustness and limited perception. It introduces DPE-MVS, a PatchMatch-based framework that leverages dual-level edge cues to guide plane construction and sampling, along with an adaptive patch size strategy. Core contributions include edge-guided non-local sampling, perception-range expansion, plane-construction optimization, and adaptive patch sizing, which together yield state-of-the-art results on ETH3D and Tanks & Temples, notably achieving the best F1 on ETH3D. The method demonstrates strong robustness to stochastic textures, maintains competitive efficiency, and lays groundwork for integrating learning-based cues to further enhance fine-detail reconstruction.

Abstract

The reconstruction of low-textured areas is a prominent research focus in multi-view stereo (MVS). In recent years, traditional MVS methods have performed exceptionally well in reconstructing low-textured areas by constructing plane models. However, these methods often encounter issues such as crossing object boundaries and limited perception ranges, which undermine the robustness of plane model construction. Building on previous work (APD-MVS), we propose the DPE-MVS method. By introducing dual-level precision edge information, including fine and coarse edges, we enhance the robustness of plane model construction, thereby improving reconstruction accuracy in low-textured areas. Furthermore, by leveraging edge information, we refine the sampling strategy in conventional PatchMatch MVS and propose an adaptive patch size adjustment approach to optimize matching cost calculation in both stochastic and low-textured areas. This additional use of edge information allows for more precise and robust matching. Our method achieves state-of-the-art performance on the ETH3D and Tanks & Temples benchmarks. Notably, our method outperforms all published methods on the ETH3D benchmark.

Dual-Level Precision Edges Guided Multi-View Stereo with Accurate Planarization

TL;DR

The paper tackles robust 3D reconstruction in low-textured regions for multi-view stereo by addressing plane-model robustness and limited perception. It introduces DPE-MVS, a PatchMatch-based framework that leverages dual-level edge cues to guide plane construction and sampling, along with an adaptive patch size strategy. Core contributions include edge-guided non-local sampling, perception-range expansion, plane-construction optimization, and adaptive patch sizing, which together yield state-of-the-art results on ETH3D and Tanks & Temples, notably achieving the best F1 on ETH3D. The method demonstrates strong robustness to stochastic textures, maintains competitive efficiency, and lays groundwork for integrating learning-based cues to further enhance fine-detail reconstruction.

Abstract

The reconstruction of low-textured areas is a prominent research focus in multi-view stereo (MVS). In recent years, traditional MVS methods have performed exceptionally well in reconstructing low-textured areas by constructing plane models. However, these methods often encounter issues such as crossing object boundaries and limited perception ranges, which undermine the robustness of plane model construction. Building on previous work (APD-MVS), we propose the DPE-MVS method. By introducing dual-level precision edge information, including fine and coarse edges, we enhance the robustness of plane model construction, thereby improving reconstruction accuracy in low-textured areas. Furthermore, by leveraging edge information, we refine the sampling strategy in conventional PatchMatch MVS and propose an adaptive patch size adjustment approach to optimize matching cost calculation in both stochastic and low-textured areas. This additional use of edge information allows for more precise and robust matching. Our method achieves state-of-the-art performance on the ETH3D and Tanks & Temples benchmarks. Notably, our method outperforms all published methods on the ETH3D benchmark.
Paper Structure (18 sections, 12 equations, 12 figures, 8 tables)

This paper contains 18 sections, 12 equations, 12 figures, 8 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Method
  4. Review of APD-MVS
  5. Overview of Our Method
  6. Extracting Edge Cues
  7. Edge Guided Non-Local Sampling
  8. Accurate Plane Model Construction
  9. Adaptive Patch Size Adjustment
  10. Experiments
  11. Datasets and Implementation Details
  12. Evaluation on MVS Benchmarks
  13. Ablation Studies
  14. Conclusion
  15. Acknowledgements
  16. ...and 3 more sections

Figures (12)

  • Figure 1: Comparison with the SOTA traditional methods and learning-based methods. Our method achieves the best F$_1$-score on ETH3D and the best recall on Tanks & Temples.
  • Figure 2: Top: depth maps for scenes crossing object boundaries. Bottom: normal maps for limited perception range. Comparison of APD-MVS (middle) and our method (right).
  • Figure 3: Comparison of plane construction between APD-MVS (middle) and our method (right), with visualizations of fine edges (top right) and coarse edges (bottom right).
  • Figure 4: Overview. DPE-MVS adopt a pyramid structure, with the two coarsest scales displayed on the right side of the figure, and the middle illustrating the details of our proposed DPE-PM. Iterations at finer scales use DPE-PM to update the depth map.
  • Figure 5: Edge Guided Non-Local Sampling: This process includes two sampling schemes: progressive non-local sampling (left) and edge-guided extended sampling (right).
  • ...and 7 more figures