Aiming at the comfort and efficiency problems of respiratory rate detection in the field of physiological health monitoring, this paper proposes a visionbased noncontact measurement method, which uses a common camera to capture the human respiratory video and enlarges the displacement of the chest and abdomen motion during breathing with Euler algorithm. Considering the influence of the position extraction accuracy of the chest and abdomen area on the accuracy of the respiratory rate detection, this paper proposes a method based on optical flow signal to extract the respiratory region. The optical flow algorithm is used to convert the chest and abdomen motion into optical flow information, which are encoded and displayed in the form of a color image. The pixel brightness sequence of the chest and abdomen breathing region is extracted to obtain respiratory waveform information, and the respiratory rate is obtained with peak detection. Finally, experiment was conducted, and the breath signals extracted using the proposed algorithm were compared with the measurement results using Embla N7000 polysomnography. The results show that the average error of respiratory rate detected with the proposed algorithm is 054 times/min, which has high accuracy.