The present master thesis is about the topic camera based heart rate monitoring focusing to automotive applications. The goal is to calculate the driver’s heart rate in the car continuously, accurate and without any distractions. With this information a more detailed picture can be acquired about the dirver’s health condition and if it’s necessarry further actions can be made.
In this thesis, the biological background of camera based heart rate monitoring is explained, the already existing methods are introduced and further developments are presented. The main task was to improve the robustness against the test subject’s motions. In oder to achieve that, the confidence values are analyzed, which are used to give information about the reliability and accuracy of the calculated heart rate. Instead of the previously used algorithm, here a new method for non contact heart rate monitoring is introduced. Furthermore a confidence value based spatiotemporal frequency tracking is implemented in order to reduce the error of the calculations and to make the results more continuous. With the evaluation of the new algorithm a camparison is made with older versions of the method in order to see how much it improved by the new components. Other methods are also introduced, like blood flow model, which can improve the algorithm even more. At the end of this master thesis a conclusion of the results is made and the further development potentials are discussed.