A non-contact method to extract human cardiac pulse rate non-invasively is more affordable and convenient compared with conventional method using contact electrode. This paper developed a video method to extract pulse rate based on smartphone. Using the front camera of smartphone to capture image frame data of human face region, defined as ROI (region of interest) for extracting PPG signal. And to suppress motion artifacts that lead to calculation error while monitoring, we need to make ROI coordinates on human face stable and accurate. So we use face detection to get the location of human face initially, then using optical flow feature tracking keep tracking the feature points in ROI. Furthermore, Kalman filter is used to stable coordinates of ROI and enhance the waveform of PPG signal. The accuracy of pulse rate detection was evaluated by compared with FDA approved vital signs monitor.