We had collected the video frame of people's index finger by the smart phone, and then extracted the human's pulse signal by analyzing the sequence images. When the original signal exists serious noise interference, the traditional heart rate detection algorithm is not applicable. Therefore, we proposed a heart rate extraction algorithm based on K-means clustering algorithm and dynamic image sequence, which can effectively extract heart rate information under the condition of strong noise. Finally, we used Bland-Altman analysis method to evaluate the accuracy of the algorithm, and the experimental result shows that it's in good agreement with the professional measuring instrument. © Published under licence by IOP Publishing Ltd.