隨著新冠肺炎疫情的嚴重影響，人們對健康的關注也日益增加，並開始注重運動。而在運動中，通常都會透過量測心率等方式來調整運動強度或者是評估運動效果，這使得運動中的心率量測日漸得到重視。目前在心率量測的方法上大致分為接觸式與非接觸式兩種，接觸式量測雖然能夠獲得穩定及準確的心率量測結果，但缺點是需要將量測裝置配戴於身上，這將會造成使用者的不適，而非接觸式量測雖然解決了該問題，但是由於運動中會產生以下狀況，如：身體晃動而造成的光線變化、因高速運動而造成的影像模糊以及運動時所造成的遮擋問題，這些問題使得非接觸式量測受到極大的挑戰。
在研究過現有的非接觸式心率量測相關文獻後，我們注意到仍有一些問題可以進行改善，主要問題整理如下:
(1) 在高強度的運動中，使用者往往難以時刻面向鏡頭或保持臉部在鏡頭所拍攝的畫面內，這在實際應用場景中可能帶來一些挑戰。
(2) 目前相對較少的研究討論了在運動中使用側臉進行心率量測的可行性，並探討這種方法的優勢。
基於上述問題，本論文基於個人電腦與網路攝影機開發了一套應用於高強度運動中的基於側臉的非接觸式心率量測系統，並透過實驗來進一步探討與改善上述問題。本論文設計了六個實驗來驗證系統的準確度，實驗內容分為無遮擋與遮擋情況下的踏步機、腳踏車、跑步機的運動實驗。在無遮擋情況下的踏步機、腳踏車、跑步機的運動實驗中10個人的平均絕對誤差(Mean Absolute Error，MAE)/均方根誤差(Root-Mean-Square Error，RMSE) 為2.020/3.056、1.201/1.735、3.549/5.292(BPM)，而SR_5/SR_10為0.94/0.98、0.99/1.00、0.83/0.91，在有遮擋的情況下MAE/RMSE為3.462/5.387、2.363/4.140、4.379/6.830(BPM)，而SR_5/SR_10為0.83/0.92、0.93/0.97、0.76/0.88。

With the severe impact of the COVID-19 pandemic, people's concern for health has been increasing, and they have started to pay more attention to exercise as a way to maintain physical well-being. During exercise, it is common to adjust the intensity or assess the effectiveness of the workout through measurements such as pulse rate. This has led to a growing emphasis on pulse rate monitoring during workout. Currently, there are generally two methods for pulse rate measurement: contact-based and non-contact-based. While contact-based measurements can provide stable and accurate results, the drawback is that they require wearing a monitoring device, which can cause discomfort for the users. On the other hand, non-contact-based measurements eliminate the need for wearing a device, but they face significant challenges due to factors such as changes in lighting caused by body movements, image blurring due to high-speed motion, and occlusion caused by physical activity.
After reviewing existing literature on non-contact pulse rate monitoring, we have identified several areas for improvement. The main issues can be summarized as follows:
(1) During high-intensity exercise, users often find it hard to consistently face the camera or keep their faces in the screen. This could pose challenges in real-world application scenarios.
(2) Currently, there is relatively limited research discussing the feasibility of using side face image for pulse rate measurement during exercise and exploring the advantages of this approach.
Based on the aforementioned issues, this thesis presents a system for non-contact pulse rate monitoring based on side face image, specifically designed for high-intensity exercise, utilizing personal computer and web camera. This thesis also conducts experiments to further investigate and address the aforementioned challenges. Six experiments were designed to validate the accuracy of the system, involving treadmill, bike, and stepper exercises under non-obstructed and obstructed conditions. In the non-obstructed conditions, the average Mean Absolute Error (MAE)/Root-Mean-Square Error (RMSE) for 10 participants were 2.020/3.056, 1.201/1.735, and 3.549/5.292 (BPM) for treadmill, bike, and stepper exercises, respectively. The SR_5/SR_10 were 0.94/0.98, 0.99/1.00, and 0.83/0.91. In the obstructed conditions, the MAE/RMSE for 10 participants were 3.462/5.387, 2.363/4.140, and 4.379/6.830 (BPM) for treadmill, bike, and stepper exercises, respectively. The SR_5/SR_10 were 0.83/0.92, 0.93/0.97, and 0.76/0.88.

[1] i運動資訊平台, "中華民國110年運動現狀調查," 2022. [Online]. Avalible: https://isports.sa.gov.tw/apps/Download.aspx?SYS=TIS&MENU_CD=M0
[2] J. H. Wilmore and D. L. Costill, Physiology of Sport and Exercise Champaign, IL: Human Kinetics, 1994.
[3] A. B. Hertzman and C. Spealman, “Observations on the finger volume pulse recorded photoelectrically,” Amer. J. Physiol., vol. 119, pp. 334-335, 1937.
[4] W. Verkruysse, L. O. Svaasand, and J. S. Nelson, “Remote plethysmographic imaging using ambient light,” Optics Express, vol. 16, no. 26, pp. 21434-21445, 2008.
[5] M. Z. Poh, D. J. McDuff, and R. W. Picard, “Non-contact automated cardiac pulse measurements using video imaging and blind source separation,” Optics Express, no. 10, pp. 10762-10774, 2010.
[6] M. Lewandowska, J. Rumiński, T. Kocejko, and J. Nowak, “Measuring pulse rate with a webcam—A noncontact method for evaluating cardiac activity,” Federated Conference on Computer Science and Information Systems, pp. 405-410, 2011.
[7] G. de Haan and V. Jeanne, “Robust pulse rate from chrominance-based rPPG,” IEEE Transactions on Biomedical Engineering, vol. 60, no. 10, pp. 2878-2886, 2013.
[8] G. de Haan and A. van Leest, “Improved motion robustness of remote-PPG by using the blood volume pulse signature,” Physiological Measurement, vol. 35, no. 9, pp. 1913-1922, 2014.
[9] W. Wang, S. Stuijk and, G. de Haan, “A novel algorithm for remote photoplethysmography: Spatial subspace rotation,” IEEE Transactions on Biomedical Engineering, vol. 63, no. 9, pp. 1974-1984, 2016.
[10] 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.
[11] C. Alvarez Casado and M. Bordallo Lopez, “Face2ppg: An unsupervised pipeline for blood volume pulse extraction from faces,” arXiv:2202.04101, 2022. [Online]. Available: https://arxiv.org/abs/2202.04101
[12] S. Kwon, J. Kim, D. Lee, and K. Park, “ROI analysis for remote photoplethysmography on facial video,” Annual International Conference of the IEEE Engineering in Medicine and Biology Society, pp. 851-862, 2015.
[13] S. Bobbia, Y. Benezeth and, J. Dubois, “Remote photoplethysmography based on implicit living skin tissue segmentation,” International Conference on Pattern Recognition, pp. 361-365, 2016.
[14] L. M. Po et al., “Block-based adaptive ROI for remote photoplethysmography,” Multimedia Tools and Applications, vol. 77, no. 6, pp. 6503-6529, 2018.
[15] W. Wang, A. C. den Brinker, and G. De Haan, “Full video pulse extraction,” Biomedical Optics Express, vol. 9, no. 8, pp. 3898-3914, 2018.
[16] K. L. Wong et al., “Optimising rPPG Signal Extraction by Exploiting Facial Surface Orientation,” IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, pp. 2165-2171, 2022.
[17] C. Lugaresi et al., “MediaPipe: A framework for building perception pipelines,” arXiv:1906.08172, 2019. [Online]. Available: https://arxiv.org/abs/1906.08172
[18] C. Zhao, M. Zhou, W. Han and, Y. Feng, “Anti-motion remote measurement of heart rate based on region proposal generation and multi-scale ROI fusion,” IEEE Transactions on Instrumentation and Measurement, vol. 71, pp. 1-13, 2022.
[19] W. Wang, A. C. den Brinker, S. Stuijk and G. de Haan, “Robust heart rate from fitness videos,” Physiological Measurement, vol. 38, no. 6, pp. 1023-1044, 2017.
[20] Y. C. Lin and Y. H. Lin, “Step count and pulse rate detection based on the contactless image measurement method,” IEEE Transactions on Multimedia, vol. 20, pp. 2223-2231, 2018.
[21] K. Xie, C. H. Fu, H. Liang, H. Hong and, X. Zhu, “Non-contact heart rate monitoring for intensive exercise based on singular spectrum analysis,” IEEE Conference on Multimedia Information Processing and Retrieval, pp. 228-233, 2019.
[22] Y. C. Chou, B. Y. Ye, H. R. Chen and, Y. H. Lin, “A real-time and non-contact pulse rate measurement system on fitness equipment,” IEEE Transactions on Instrumentation and Measurement, vol. 71, pp. 1-11, 2022.
[23] Y. Napolean et al., “Heart rate estimation in intense exercise videos,” IEEE International Conference on Image Processing, Bordeaux, France, pp. 3933-3937, 2022.
[24] R. Spetlik, V. Franc, J. Cech and, J. Matas, “Visual heart rate estimation with convolutional neural network,” British Machine Vision Conference, pp. 3-6, 2018.
[25] X. Niu, S. Shan, H. Han and, X. Chen, “Rhythmnet: End-to-end heart rate estimation from face via spatial-temporal representation,” IEEE Transactions on Image Processing, vol. 29, pp. 2409-2423, 2019.
[26] 范哲嘉, “基於線性回歸分析之強健的非接觸式心率量測系統在運動器材的應用,” 國立臺灣科技大學, 2022.
[27] R. Achanta et al., “SLIC superpixels compared to state-of-the-art superpixel methods,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 34, no. 11, pp. 2274-2282, 2012.
[28] OpenCV, “Face detector.” [Online]. Available: https://github.com/opencv/opencv/tree/master/samples/dnn/face_detector
[29] X. Ren et al., “Facial geometric detail recovery via implicit representation,” IEEE 17th International Conference on Automatic Face and Gesture Recognitio, Waikoloa Beach, HI, USA, pp. 1-8, 2023.
[30] Dlib, “Dlib C++ library.” [Online]. Available: https://github.com/davisking/dlib
[31] X. Ren and J. Malik, “Learning a classification model for segmentation,” International Conference on Computer Vision, vol. 1, pp. 10–17, 2003.
[32] C. Sagonas, G. Tzimiropoulos, S. Zafeiriou and, M. Pantic, “300 faces in-the-wild challenge: The first facial landmark localization challenge,” IEEE International Conference on Computer Vision Workshops, pp. 397-403, 2013.
[33] A. Kim, “Fast-slic.” [Online]. Available: https://github.com/Algy/fast-slic
[34] J. F. Henriques, R. Caseiro, P. Martins, and J. Batista, “High-speed tracking with kernelized correlation filters,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 37, no. 3, pp. 583-596, 2015.
[35] “Polar H7 heart rate sensor.” [Online]. Available: https://support.polar.com/e_manuals/H7_Heart_Rate_Sensor/Polar_H7_Heart_Rate_Sensor_accessory_manual_English.pdf
[36] “Vernier Go Direct EKG sensor.” [Online]. Available: https://www.vernier.com/product/go-direct-ekg-sensor/
[37] B. F. Wu et al., "Motion Resistant Image-Photoplethysmography Based on Spectral Peak Tracking Algorithm," IEEE Access, vol. 6, pp. 21621-21634, 2018.