EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

SPTNet: SuperPoint Tracking Network for Visual SLAM

Min Pang, Jichao Jiao iD, Yingjian Zhang

Sensors · 21.07.2026 · 26 (14) · 4612

DOI: 10.3390/s26144612 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, online erschienen am 21.07.2026. Verfasst von Min Pang und 2 weiteren Autor:innen. Beteiligte Einrichtung: School of Electronic Engineering, Beijing University of Posts and Telecommunications, No. 10 Xitucheng Road, Haidian District, Beijing 100876, China. Die Arbeit nennt 29 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Robust image-to-image correspondence is a fundamental challenge for camera-based Visual Simultaneous Localization and Mapping (SLAM). Conventional approaches primarily rely on isolated local feature matching or optical flow prediction, which often suffer from limited robustness under large parallax, high computational overhead, and cumulative drift errors during long-sequence tracking. To address these limitations, we propose SPTNet (SuperPoint Tracking Network), an efficient multi-task neural network that tightly couples feature detection, description, and dense optical flow prediction within a unified architecture. The fundamental innovation of SPTNet is a Hybrid Tracking Module (HTM) governed by a novel Predictor-Corrector mechanism. Specifically, the dense optical flow field acts as a temporal prior to constrain the descriptor matching search space, while the descriptors act as a correction signal, eliminating flow-induced drift at each frame through spatially constrained Sinkhorn optimization. This synergy enables efficient feature reuse via a shared backbone, minimizing redundant computation. Comprehensive experiments on indoor and outdoor datasets demonstrate that SPTNet attains a false matching rate as low as 1.8% at a 5-pixel threshold on HPatches, substantially reduces cumulative drift on long-sequence SLAM benchmarks, and maintains a high execution speed of 35 FPS on standard GPUs, demonstrating a highly compact footprint advantageous for prospective embedded robotic deployment.

Schlagwörter

Stichwörter der Autor:innen

SLAMdeep learningfeature matchingneural networksoptical flow

Autor:innen und Institutionen

  • School of Electronic Engineering, Beijing University of Posts and Telecommunications, No. 10 Xitucheng Road, Haidian District, Beijing 100876, China.
  • China Research Institute of Radiowave Propagation, No. 33 Xianshan East Road, Chengyang District, Qingdao 266107, China.

ORCID-Profile: Jichao Jiao

Zitieren

APA

Pang, M., Jiao, J., & Zhang, Y. (2026). SPTNet: SuperPoint Tracking Network for Visual SLAM. Sensors, 26(14), 4612. https://doi.org/10.3390/s26144612

Vancouver

Pang M, Jiao J, Zhang Y. SPTNet: SuperPoint Tracking Network for Visual SLAM. Sensors. 2026;26(14):4612. doi:10.3390/s26144612

BibTeX
@article{Pang2026,
  author = {Pang, Min and Jiao, Jichao and Zhang, Yingjian},
  title = {SPTNet: SuperPoint Tracking Network for Visual SLAM},
  journal = {Sensors},
  year = {2026},
  volume = {26},
  number = {14},
  pages = {4612},
  doi = {10.3390/s26144612}
}
Kostenlos starten →