Deteksi Detak Jantung Menggunakan Kamera Dual Modalitas NIR-RGB Pada Aktivitas Olahraga Lari

Naufal, Arvin Daffa (2025) Deteksi Detak Jantung Menggunakan Kamera Dual Modalitas NIR-RGB Pada Aktivitas Olahraga Lari. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5026211014-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Kesehatan jantung sangat penting, dengan aritmia membawa risiko serius seperti stroke dan kematian mendadak. Metode tradisional, seperti penghitungan denyut nadi manual atau EKG, terbatas dalam efisiensi dan kenyamanan. Penelitian ini mengembangkan sistem remote photoplethysmography (rPPG) berbasis kamera dual-modalitas (RGB dan NIR) untuk deteksi detak jantung non-kontak saat olahraga lari. Sistem ini menggunakan metode pemrosesan sinyal yang terdiri dari ROI Localization, ekstraksi sinyal RGB/NIR, modified amplitude selective filtering, wavelet decomposition, dan independent component analysis (ICA). Deteksi detak jantung dilakukan melalui transformasi Fourier terhadap sinyal terpilih untuk menentukan puncak frekuensi dominan dalam rentang 0,7–3,5 Hz. Tiga mode input diuji: Mode 0 (RGB), Mode 1 (RGB + NIR), dan Mode 2 (CIELab a,b + NIR), dan dibandingkan terhadap data EKG sebagai ground truth. Hasil menunjukkan bahwa Mode 0 (RGB) efektif untuk kondisi minim gerakan, seperti sebelum lari, dengan RMSE terendah 27,17 bpm. Namun, saat terjadi pergerakan intensif, Mode 1 (RGB + NIR) menunjukkan akurasi terbaik hingga 83%, dengan performa stabil meskipun terdapat gangguan pencahayaan dan artefak gerakan. Secara keseluruhan, sistem rPPG mampu mengikuti detak jantung secara real-time, dengan sinyal NIR menjadi komponen penting pada aktivitas berat, sementara RGB cukup untuk kondisi statis.
====================================================================================================================================================================================
Heart health is crucial, with arrhythmias carrying serious risks such as stroke and sudden death. Traditional methods, such as manual pulse counting or ECG, are limited in efficiency and convenience. This study developed a dual-modality (RGB and NIR) camera-based remote photoplethysmography (rPPG) system for non-contact heart rate detection during running. The system utilizes a signal processing consisting of ROI localization, RGB/NIR signal extraction, modified amplitude selective filtering, wavelet decomposition, and independent component analysis (ICA). Heart rate detection is performed by Fourier transforming selected signals to determine the dominant frequency peaks in the 0.7–3.5 Hz range. Three input modes were tested: Mode 0 (RGB), Mode 1 (RGB + NIR), and Mode 2 (CIELab a,b + NIR), and compared against ECG data as ground truth. Results showed that Mode 0 (RGB) was effective for low-motion conditions, such as before running, with the lowest RMSE of 27.17 bpm. However, during intense movement, Mode 1 (RGB + NIR) demonstrated the best accuracy of up to 83%, with stable performance despite lighting interference and motion artifacts. Overall, the rPPG system is capable of tracking heart rate in real time, with the NIR signal being a crucial component during vigorous activity, while RGB is sufficient for static conditions.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Deteksi detak jantung, rPPG, lari, kamera NIR-RGB, Heart rate detection, rPPG, running, NIR-RGB camera.
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5102.9 Signal processing.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 57201-(S1) Undergraduate Thesis
Depositing User:	Arvin Daffa Naufal
Date Deposited:	25 Jul 2025 06:18
Last Modified:	25 Jul 2025 06:18
URI:	http://repository.its.ac.id/id/eprint/121525

Actions (login required)

View Item