Abstract: A new method to measure the respiration based on fast face detection is proposed. The camera captures the variation of chest contour by human respiratory, human respiratory wave signal extracted from this variation can be used to calculate real-time respiratory rate. The system detects the human face position in the first image obtained by the camera through the color-based face detecting method and gets the chest position according the priori knowledge of body geometry, then gets gray values of the chest area in each frame, processes the data by wavelet de-noising and other methods to obtain real-time respiration rate. The actual measurement experiments and comparative experiments show that the system is accuracy and practical.

Key words: respiratory signal, face detection, image processing, wavelet transform

摘要： 提出一种基于快速人脸检测的呼吸测量方法，通过普通摄像头捕捉人体呼吸所带来的胸口轮廓的变化规律，提取人体呼吸波信号，计算实时呼吸率。系统采用基于肤色的人脸检测方法，可以快速检测出摄像头获取的图像中人脸位置，根据人脸和胸口的一般化距离，实现胸口定位，获得此后所获取的每一帧图像中胸口区域的灰度值，对数据进行小波去噪等一系列处理，通过算法获得实时呼吸率。通过实际测量实验和对比实验表明，该系统具有很好的准确性和实用价值。

关键词: 呼吸信号, 人脸检测, 图像处理, 小波变换