A motion-robust contactless photoplethysmography using chrominance and adaptive filtering

Recently, the camera-based photoplethysmography (PPG) techniques have been proposed for heart rate measurement due to the contactless and low-cost features. However, the previous works showed limited success for moving subjects, especially when the signal of skin color changes rendered by vigorous motion is in the same frequency band as that by blood volume variations. This paper proposed a motion robust, contactless method for measuring heart rate when the human subject is exercising on some fitness machines, e.g. treadmill or stepper. The proposed method combines the chrominance-based method with adaptive filtering to make the heart rate measurement robust to vigorous motions on fitness machine while estimating the heart rate in frequency domain. To demonstrate the performance of heart rate measurement, we taped a video sequence of treadmill exercise with various motion themes. As shown in experimental results on the video sequence, our method can significantly improve the measurement accuracy from 61% to 90%, comparing to the previous chrominance-based method.