Skin Reflection Angle Useful for Region of Interest Selection in Camera-based Heart Rate Estimation?
Year: 2024
Type: conference-paper
Abstract: Photoplethysmography Imaging (PPGI) is considered a promising non-contact alternative for existing methods of heart rate (HR) and even blood pressure estimation.A critical processing step in PPGI involves selecting the pixels with the strongest pulse signal from a camera video, which becomes particularly challenging in real-world scenarios involving movement and changing illumination.Current pixel selection methods such as facial landmark-based region selection and skin segmentation overlook scene geometry and illumination conditions.This study evaluates whether the angle between the camera sensor and the skin surface (skin-to-camera angle) could enhance pulse estimation accuracy.The skin-to-camera angle is calculated using 3D-coordinate estimates, and thresholds ranging from 15° to 90° are applied to reject sub-regions based on their angle.The effectiveness of this angle-based selection method is tested on six publicly available datasets containing video data from 288 subjects.Results show that for the CHROM algorithm, angle thresholds of 30° to 50° marginally improve heart rate estimation accuracy by 0.1 BPM, but most algorithms do not benefit significantly from this approach.Consequently, due to the added computational complexity, the angle is unlikely to be a substantial confounder in PPGI signal quality.
Source: Computing in cardiology
Institution: Technische Universität Darmstadt
Language: English
Subfield: Biomedical Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W4405435642