Skip to Main content Skip to Navigation

Semi-blind Source Extraction Methods: Application to the measurement of non-contact physiological signs

Abstract : Non-contact physiological measurements are highly desirable in many biomedical fields such as diagnosis of infants, geriartic patients, patients with extreme physical trauma, and fitness and well-being. Remote photoplethysmography is increasingly being used for non-contact measurement of heart rate from videos which is one of the most common biomedical property required for most medical diagnosis. One of the common techniques for performing remote photoplethysmography involves using Blind Source Separation (BSS) methods to extract the cardiac signal from video data. In this context, the objective of this thesis is to develop different methods in the field of extraction and separation of sources by improving upon traditional BSS methods. These novel semiblind source extraction methods are integrated with biophysical constraints, and applied to the application of remote photoplethysmography measurement. In addition, one of these methods is extended to measure the spatial distribution of photoplethysmographic signals of the skin. Remote photoplethysmography aims to measure biophysical parameters such as heart rate and heart rate variability by quantifying the periodic changes in skin color due to the rhythmic beating of the heart. These changes manifest in the image data obtained from simple video cameras, which is processed to generate a temporal signal representing the cardiac signal. We have improved existing methods by incorporating the ubiquitous property of quasiperiodicity of biophysical signals such as cardiac and neurological signals. Quasi-periodic signals have higher autocorrelation than non-periodic signals. This observation was combined with independent component analysis techniques and Generalized Eigenvalue Decomposition (GEVD) to develop semiblind source extraction methods.
Complete list of metadatas

Cited literature [128 references]  Display  Hide  Download
Contributor : Richard Macwan <>
Submitted on : Tuesday, March 26, 2019 - 11:57:27 PM
Last modification on : Friday, July 17, 2020 - 2:54:07 PM
Long-term archiving on: : Thursday, June 27, 2019 - 6:24:45 PM


PhDThesis-Richard Macwan.pdf
Files produced by the author(s)


  • HAL Id : tel-02080653, version 1


Richard Macwan. Semi-blind Source Extraction Methods: Application to the measurement of non-contact physiological signs. Computer Vision and Pattern Recognition [cs.CV]. Université de Bourgogne - Franche Comté, 2018. English. ⟨tel-02080653⟩



Record views


Files downloads