隨著預防醫學與遠距醫療的興起，非接觸式生理訊號監測技術已成為研究熱點。本研究提出一套基於遠端光體積變化描記圖（remote Photoplethysmography, rPPG）技術之非接觸式血壓估測系統，專注於分析手部（手掌與手背）區域之微弱光學變化。系統透過一般視訊鏡頭（Webcam）擷取影像，並應用平面正交投影（Plane-Orthogonal-to-Skin, POS）演算法與帶通濾波器提取血液容積脈搏訊號（Blood Volume Pulse, BVP），建立可於一般室內環境運作之影像式生理訊號處理流程。
本研究採用混合特徵提取策略，結合傳統生理統計特徵（如心率、脈搏振幅與心跳間隔變異性）以及一維卷積類神經網路自編碼器（1D-CNN Autoencoder）所學習之深度波形潛在特徵，以提升模型對非線性脈搏形態與個體差異之表達能力。在模型建構方面，針對收縮壓（SBP）與舒張壓（DBP）分別訓練隨機森林回歸模型，並導入 Ridge 回歸校正機制以降低系統性預測偏差。
實驗結果顯示，於跨受試者(20人)驗證情境下，結合混合特徵與個人化校正策略可顯著提升血壓估測準確度。其中，SBP 平均絕對誤差（MAE）由約 14–16 mmHg 降低至約 5–6 mmHg，DBP MAE 則由約 8–10 mmHg 降低至約 2.7–3.1 mmHg，顯示深度潛在特徵與校正機制對預測穩定性具關鍵貢獻。與未導入個人化調整之模型相比，本方法於跨個體應用情境下展現出顯著誤差降低效果。
此外，相較於臉部量測方式，手部 rPPG 具備較低隱私風險與較高情境可用性，適合應用於日常居家與辦公環境之長期監測需求。整體而言，本研究所提出之混合特徵非接觸式血壓估測方法，在一般硬體設備下即可達成具臨床潛力之預測效能，為非接觸式生理監測系統提供一項可行解決方案。

With the rise of preventive medicine and telehealth, contactless physiological monitoring technologies have become an active research focus. This study proposes a contactless blood pressure (BP) estimation system based on remote photoplethysmography (rPPG) to analyze subtle optical variations in hand reg-ions, including the palm and the back of the hand. Video sequences are captured using a standard web-cam, and Blood Volume Pulse (BVP) signals are extracted using the Plane-Orthogonal-to-Skin (POS) algorithm combined with a bandpass filter.

A hybrid feature extraction strategy integrates conventional physiological statistical features with mor- phological waveform representations learned by a one-dimensional convolutional neural network autoencoder (1D-CNN Autoencoder). Separate Random Forest regression models are trained for systol-ic blood pressure (SBP) and diastolic blood pressure (DBP), followed by a Ridge regression-based cali-bration mechanism to reduce systematic prediction bias. Cross-subject performance is evaluated using Leave-One-Subject-Out (LOSO) validation.

Experimental results show that the proposed framework significantly improves BP estimation accuracy, reducing SBP mean absolute error (MAE) from approximately 14–16 mmHg to 5–6 mmHg and DB-P MAE from approximately 8–10 mmHg to 2.7–3.1 mmHg. These findings demonstrate that deep waveform representations and personalization-based calibration effectively mitigate inter-subject varia-bility. Compared to face-based measurements, hand-based rPPG provides enhanced privacy preservati-on and practical applicability for long-term contactless BP monitoring.

[1] J.-R. Hu, G. Martin, S. Iyengar, L. C. Kovell, T. B. Plante, N. van Helmond, R. A. Dart, T. M. Brady, R.-A. N. Turkson-Ocran, and S. P. Juraschek, “Validating cuffless continuous blood pressure monitoring devices,” 2023. PMCID: PMC9971997; PMID: 36865583.
[2] R. Mukkamala, M. Yavarimanesh, K. Natarajan, J.-O. Hahn, K. G. Kyriakoulis, A. P. Avolio, and G. S. Stergiou, “Evaluation of the accuracy of cuffless blood pressure measurement devices: Challenges and proposals,”Hypertension, vol. 78, no. 5, 2021, doi: 10.1161/HYPERTENSIONAHA.121.17747.
[3] T. Tamura, Yuka Maeda, M. Sekine, Masaki Yoshida, "Wearable Photoplethysmographic Sensors—Past and Present",　https://www.semanticscholar.org/reader/ca04bcc2fdd9aed2b04e1f17c6bbb62f44dd3668, (Apr. 1, 2014),（Last Access 02/03/2026）
[4] H. Gesche, D. Grosskurth, G. Küchler, and A. Patzak, “Continuous blood pressure measurement by using the pulse transit time: Comparison to a cuff-based method,” European Journal of Applied Physiology, vol. 112, pp. 309–315, 2012.
[5] M. Cao, T. Burton, G. Saiko, and A. Douplik, “Remote photoplethysmography with a high-speed camera reveals temporal and amplitude differences between glabrous and non-glabrous skin,” Sensors, 2023. PMCID: PMC9866593; PMID: 36679411.
[6] J. B. Cohen, R. L. Byfield, S. T. Hardy, S. P. Juraschek, N. H. Miller, R. Mukkamala, D. S. Picone, R. H. Thiele, E. Yang, and T. M. Brady, “Cuffless devices for the measurement of blood pressure: A scientific statement from the American Heart Association,”Hypertension, 2024, doi: 10.1161/HYP.0000000000000254.

[7] K. L. Wong, J. W. Chin, T. T. Chan, I. Odinaev, K. Tianqu and R. H. Y. So, “Optimising rPPG signal extraction by exploiting facial surface orientation,” in Proc. CVPR Workshop, 2022, pp. 0–0, doi: 10.1109/CVPRW56347.2022.
[8] J. Allen, “Photoplethysmography and its application in clinical physiological measurement,” Physiological Measurement, vol. 28, no. 3, pp. R1–R39, 2007, doi: 10.1088/0967-3334/28/3/R01. PMID: 17322588.
[9] W. Wang, A. C. den Brinker, S. Stuijk, and G. de Haan, “Algorithmic principles of remote PPG,” IEEE Transactions on Biomedical Engineering, vol. 64, no. 7, pp. 1479–1491, 2017.
[10] A. Atyabi and N. M. Ahmad, “Deep learning based feature extraction for blood pressure estimation using photoplethysmography,” Biomedical Signal Processing and Control, vol. 58, 2020, Art. no. 101831.
[11] F. Schrumpf, P. Frenzel, C. Aust, G. Osterhoff, and M. Fuchs, “Assessment of Non-Invasive blood pressure prediction from PPG and rPPG signals using deep learning,” Sensors, vol. 21, no. 18, p. 6022, 2021, doi: 10.3390/s21186022.
[12] X. Niu, H. Han, S. Shan, and X. Chen, “A comparative survey of methods for remote heart rate detection from frontal face videos,” Frontiers in Bioengineering and Biotechnology, vol. 6, 2018, Art. no. 33, doi: 10.3389/fbioe.2018.00033.
[13] G. de Haan and A. van Leest, “Improved motion robustness of remote-PPG by using the blood volume pulse signature,” Physiological Measurement, vol. 35, pp. 1913–1926, 2014.
[14] A. Moco, S. Stuijk, M. van Gastel, and G. de Haan, “Impairing factors in remote-PPG pulse transit time measurements on the face,” in Proc. IEEE/CVF Conf. Computer Vision and Pattern Recognition Workshops (CVPRW), 2018.
[15] G. Zhang, M. Gao, D. Xu, N. B. Olivier, and R. Mukkamala, “Pulse arrival time is not an adequate surrogate for pulse transit time as a marker of blood pressure,” Journal of Applied Physiology, vol. 111, pp. 1681–1686, 2011.
[16] G. S. Stergiou, A. Kollias, C. Destounis, and G. T. O’Brien, “A universal standard for the validation of blood pressure measuring devices: Association for the Advancement of Medical Instrumentation/European Society of Hypertension/International Organization for Standardization (AAMI/ESH/ISO) Collaboration Statement,” Hypertension, vol. 71, pp. 368–374, 2018.
[17] 黃信豪，手心與額頭 rPPG 非接觸式血壓推估研究，成功大學工程科學系碩士論文，2025。
[18] 蔡明璋,應用毫米波雷達偵測胸腔位移經由信號處理與機器學習推算呼吸、心跳與血壓，成功大學工程科學系碩士論文， 2024。