Vehicle Driver of Real-time Heart Rate Monitoring with Camera-Based Face Recognition

A novel framework has been proposed a remote heart rate (HR) monitoring based on facial recognition and real-time video streaming from a camera. This framework also creates a stable model for long-term HR monitoring and enhances facial identification accuracy while the object is moving. The camera shoots video at a rate of 10 frames per second. Face identification and object tracking are integrated to generate a collection of rectangular faces, which are then sampled in grooves for color changes. The signal related to the HR is represented by the average of the colors inside specified areas of interest (ROI) on faces. HR is derived from differences in face skin color produced by the flow of blood. The feature extraction of signal processing methods such as Fast Fourier Transform (FFT) and Independent Component Analysis (ICA) were used to the color channels in video recordings in order to determine the HR rate derived from the face area. The results show a high level of sensitivity, predictability, accuracy, and mistake detection. The procedure is very exact and consistent.