University of Oulu

Heart rates estimation using rPPG methods in challenging imaging conditions

Saved in:
Author: Nguyen, Nhi1
Organizations: 1University of Oulu, Faculty of Information Technology and Electrical Engineering, Computer Science
Format: ebook
Version: published version
Access: open
Online Access: PDF Full Text (PDF, 2.2 MB)
Pages: 61
Persistent link:
Language: English
Published: Oulu : N. Nguyen, 2023
Publish Date: 2023-05-16
Thesis type: Master's thesis (tech)
Tutor: Bordallo Lopez, Miguel
Nguyen, Le
Reviewer: Nguyen, Le
Bordallo Lopez, Miguel


The cardiovascular system plays a crucial role in maintaining the body’s equilibrium by regulating blood flow and oxygen supply to different organs and tissues. While contact-based techniques like electrocardiography and photoplethysmography are commonly used in healthcare and clinical monitoring, they are not practical for everyday use due to their skin contact requirements. Therefore, non-contact alternatives like remote photoplethysmography (rPPG) have gained significant attention in recent years. However, extracting accurate heart rate information from rPPG signals under challenging imaging conditions, such as image degradation and occlusion, remains a significant challenge. Therefore, this thesis aims to investigate the effectiveness of rPPG methods in extracting heart rate information from rPPG signals in these imaging conditions. It evaluates the effectiveness of both traditional rPPG approaches and rPPG pre-trained deep learning models in the presence of real-world image transformations, such as occlusion of the faces by sunglasses or facemasks, as well as image degradation caused by noise artifacts and motion blur. The study also explores various image restoration techniques to enhance the performance of the selected rPPG methods and experiments with various fine-tuning methods of the best-performing pre-trained model. The research was conducted on three databases, namely UBFC-rPPG, UCLA-rPPG, and UBFC-Phys, and includes comprehensive experiments. The results of this study offer valuable insights into the efficacy of rPPG in practical scenarios and its potential as a non-contact alternative to traditional cardiovascular monitoring techniques.

see all

Copyright information: © Nhi Nguyen, 2023. This publication is copyrighted. You may download, display and print it for your own personal use. Commercial use is prohibited.