基于ICEEMD的人脸视频心率检测

Heart rate measurement from face video based on ICEEMD

基于BCG(心冲击)原理,应用ICEEMD(改进的完备整体经验模态分解),实现人脸视频的心率检测。首先,在人脸区域选取任意大小的感兴趣区域(ROI),将其中心点作为特征点进行跟踪,取该点在垂直方向的运动轨迹作为原始信号。然后,采用ICEEMD将原始信号分解为一系列的模态,选择主要频率在0. 7~3 Hz(正常心率范围)的模态重构作为脉搏信号。最后使用峰值计数或者功率谱的方法计算心率。通过与指夹式脉搏计对比,在测试者中获得的视频心率具有较高准确率。

Based on the principle of Ballistocardiogram( BCG),it applies ICEEMD to realize heart rate measurement of face video. Firstly,The center point of a region of interest( ROI) of any size selected in the face region is tracked as a feature point. And the motion trajectory of this point in the vertical direction is taken as the original signal which is decomposed into a series of modes using ICEEMD. Then,the modes whose main frequency within the range of 0. 7 Hz to 3 Hz( normal heart rate range) is selected for reconstruction as a pulse signal. Finally,heart rate can be calculated from the peaks or power spectrum of pulse signal. Compared with a finger-clip pulse meter,the heart rate obtained from the methodology presented has a higher accuracy.