Colour based heart rate detection

Abstract

In this thesis, assigned by Plantronics®, the analysis and implementation of a colour based heart rate sensor will be discussed, which can be integrated in a sports headset. Two implementations are discussed: an RGB (Red, Green and Blue) sensor and a Light Dependent Resistor (LDR). The former setup identifies colour variations while the latter detects intensity changes in the whole spectrum. The data acquisition is realised by means of an Arduino as Analog-to-Digital Converter (ADC), which sends the information to a computer via a continuous data link. The processing of the received data is then performed inMATLAB. This document reports the progress and achieved results of the project. At first, the theory behind the measured parameters will be discussed. A concept is designed to proof the principle of measuring colour variations by an RGB sensor and LDR. The next step is constructing a setup that is able to detect skin discolouration, caused by the pulsating blood flow. This periodically changing skin colour can be characterised by amplitude differences. To distinguish the heart rate from noise artefacts, a peak detection algorithm and a Fast Fourier Transformation (FFT) are used. The results of all techniques are subsequently discussed. A conclusion is drawn about these results, after which future recommendations are discussed. Both the RGB and LDR concept were able to detect colour variations. Although the results of the simulated RGB setup showed promising results, actual skin discolouration measurements did not have consistent results. The LDR configuration did however identify colour intensity changes, resulting in a measurable heartbeat.