A Novel Stereo Matching Algorithm for Tunnel Reconstruction Based on Improved Census‐SAD Fusion and Gradient Information
Journal of Sensors · 2026-01 · 2026 (1)
DOI: 10.1155/js/2181507 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Sensors, online erschienen am 13.07.2026. Verfasst von Fei Geng und 3 weiteren Autor:innen. Die Arbeit nennt 22 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
The presence of weak texture and salt‐and‐pepper noise in tunnel images poses significant challenges to the accuracy of stereo matching. Existing methods have prominent drawbacks: Fixed‐window designs fail to balance weak‐texture smoothing and texture‐rich detail preservation, over‐reliance on central pixels leads to poor noise resistance, insufficient fusion of relative grayscale relationships and absolute grayscale values leading to inferior matching accuracy, and poor adaptability to tunnel‐specific harsh environments causes high mismatch rates. To address these issues, this paper introduces a novel stereo matching algorithm that enhances and combines the traditional Census transform and sum of absolute difference (SAD) algorithm. The traditional Census transform is enhanced by adopting adaptive windows to better process images with prevalent weak textures. Additionally, the encoding method is upgraded from single‐bit to a new approach utilizing three two‐bit combinations, which effectively mitigates the influence of noise. Then, the improved SAD is integrated with the enhanced Census, leveraging both the specific values and relative relationships between pixels. Experimental results indicate that the proposed algorithm outperforms classic mainstream stereo matching algorithms including semiglobal matching (SGM) and AD‐Census in denoising and weak‐texture processing, achieving a tunnel reconstruction error rate of less than 3% and satisfying practical engineering requirements. The contributions of this study are fourfold: (1) proposing an adaptive window approach for the Census transform to address fixed‐window limitations; (2) developing an enhanced Census encoding technique to reduce central pixel reliance and improve noise robustness; (3) integrating the improved Census and SAD to balance processing capabilities for different texture regions; and (4) validating the algorithm through comprehensive experiments, providing evidence of its superiority and advancing stereo matching applications in harsh tunnel environments.
Zitieren
APA
Geng, F., Sui, X., Wang, H., & Xie, C. (2026). A Novel Stereo Matching Algorithm for Tunnel Reconstruction Based on Improved Census‐SAD Fusion and Gradient Information. Journal of Sensors, 2026(1). https://doi.org/10.1155/js/2181507
Vancouver
Geng F, Sui X, Wang H, Xie C. A Novel Stereo Matching Algorithm for Tunnel Reconstruction Based on Improved Census‐SAD Fusion and Gradient Information. Journal of Sensors. 2026;2026(1). doi:10.1155/js/2181507
BibTeX
@article{Geng2026,
author = {Geng, Fei and Sui, Xueshuai and Wang, Hong and Xie, Changze},
title = {A Novel Stereo Matching Algorithm for Tunnel Reconstruction Based on Improved Census‐SAD Fusion and Gradient Information},
journal = {Journal of Sensors},
year = {2026},
volume = {2026},
number = {1},
doi = {10.1155/js/2181507}
}
EUVIMED