EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

Adaptive Pulse Extraction Via Filter Bank Common Spatial Pattern for Remote Photoplethysmography

Jukyung Lee, Jihwan Woo, Hyosung Joo

IEEE Access · 2026 · 14 · 106787-106803

DOI: 10.1109/access.2026.3712776 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von Jukyung Lee und 2 weiteren Autor:innen. Die Arbeit nennt 32 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0). Gefördert von: National Research Foundation of Korea, Ministry of Science and Information and Communication Technology.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Remote Photoplethysmography (rPPG) is a video-based, non-contact heart rate estimation technology. Since it enables the detection of physiological signals without attaching sensors to the skin, it is widely utilized in healthcare and telemedicine applications. In recent years, research on pulse extraction algorithms for extracting pulse signals from facial videos has been actively conducted. A major challenge in rPPG is the low signal-to-noise ratio, where faint physiological signals are easily overwhelmed by artifacts from illumination changes or head movements. Consequently, the ability to separate and extract pulse signals accurately and cleanly is regarded as a critical determinant of rPPG performance. To address these challenges, this paper proposes an adaptive pulse extraction algorithm termed Filter Bank Common Spatial Pattern (FBCSP). FBCSP learns Common Spatial Pattern filters that maximize the variance of the pulse signal while minimizing the variance of noise from the input signals, thereby effectively extracting clean signal components with suppressed noise. To validate the effectiveness of the proposed approach, we compared its performance against existing pulse extraction algorithms, including Plane Orthogonal to Skin (POS) and CHROM. Experimental results demonstrate that the proposed algorithm reduced the Mean Absolute Error (MAE) by an average of 45% compared to the widely used POS, achieving high accuracy with a final MAE of 2.35 bpm. Ultimately, the FBCSP framework extracts optimal signals by dynamically adapting to user-specific signal characteristics and varying noise environments, thereby demonstrating its capability to deliver personalized pulse signals tailored to individual spatial signal distributions and noise conditions over conventional static algorithms.

Förderung

  • National Research Foundation of Korea
  • Ministry of Science and Information and Communication Technology

Zitieren

APA

Lee, J., Woo, J., & Joo, H. (2026). Adaptive Pulse Extraction Via Filter Bank Common Spatial Pattern for Remote Photoplethysmography. IEEE Access, 14, 106787-106803. https://doi.org/10.1109/access.2026.3712776

Vancouver

Lee J, Woo J, Joo H. Adaptive Pulse Extraction Via Filter Bank Common Spatial Pattern for Remote Photoplethysmography. IEEE Access. 2026;14:106787-106803. doi:10.1109/access.2026.3712776

BibTeX
@article{Lee2026,
  author = {Lee, Jukyung and Woo, Jihwan and Joo, Hyosung},
  title = {Adaptive Pulse Extraction Via Filter Bank Common Spatial Pattern for Remote Photoplethysmography},
  journal = {IEEE Access},
  year = {2026},
  volume = {14},
  pages = {106787--106803},
  doi = {10.1109/access.2026.3712776}
}
Kostenlos starten →