EUVIMEDEUVIMED entdecken
Crossref

Generative Model-Based Matching for Infrared and RGB Camera Images

Wei Tian, Xinning Chu, Luoxin Ye

IEEE Transactions on Consumer Electronics · 2025-02 · 71 (1) · 102-114

DOI: 10.1109/tce.2024.3510564 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Consumer Electronics, 2025-02. Verfasst von Wei Tian und 2 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 50 Literaturangaben. Gefördert von: Natural Science Foundation of Chongqing, National Key Research and Development Program of China, Shanghai Municipal Science and Shanghai Automotive Industry Science and Technology Development Foundation.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Perception in light-varying scenarios is a major challenge for visual sensors, especially RGB cameras, installed on intelligent vehicles. Leveraging the illumination invariance of infrared images, a perception system based on the fusion of both infrared and RGB camera images is feasible to enhance the perceptive performance. However, due to the significant difference between the spectral modalities, the matching between infrared and RGB images is still challenging. To address such a problem, this paper proposes a generative model-based cross-spectral image matching approach based on the generative multi-modality conversion and the fusion of point and line features. In this approach, we minimize the spectral modality difference by transferring visible images to pseudo-infrared images, and explore on the optimal loss functions to learn to preserve image similarities. We also design an effective feature extraction and matching strategy for both keypoint and line segment features. Furthermore, we propose an accurately aligned cross-spectral image dataset as a complement to existing public datasets to validate the efficacy of proposed approach. The experimental results demonstrate that our approach is appropriate for cross-spectral image matching under different lighting conditions with significantly accuracy gain over existing methods.

Förderung

  • Natural Science Foundation of Chongqing
  • National Key Research and Development Program of China
  • Shanghai Municipal Science and Shanghai Automotive Industry Science and Technology Development Foundation
  • Original Research Project of Tongji University

Zitieren

APA

Tian, W., Chu, X., & Ye, L. (2025). Generative Model-Based Matching for Infrared and RGB Camera Images. IEEE Transactions on Consumer Electronics, 71(1), 102-114. https://doi.org/10.1109/tce.2024.3510564

Vancouver

Tian W, Chu X, Ye L. Generative Model-Based Matching for Infrared and RGB Camera Images. IEEE Transactions on Consumer Electronics. 2025;71(1):102-114. doi:10.1109/tce.2024.3510564

BibTeX
@article{Tian2025,
  author = {Tian, Wei and Chu, Xinning and Ye, Luoxin},
  title = {Generative Model-Based Matching for Infrared and RGB Camera Images},
  journal = {IEEE Transactions on Consumer Electronics},
  year = {2025},
  volume = {71},
  number = {1},
  pages = {102--114},
  doi = {10.1109/tce.2024.3510564}
}
Kostenlos starten →