抗相机运动干扰的心率检测算法研究

针对现有的基于视频的心率检测方法中没有考虑相机晃动产生的运动干扰导致血液容积脉搏（blood volume pulse，BVP）波信号提取不准确、检测精度低的问题，文章提出一种抗相机运动干扰的心率检测算法，利用对晃动视频的稳像处理，获取稳定的面部视频用于心率检测。首先通过加速稳健特征（speeded up robust features，SURF）算法选取特征点，进行特征匹配并求解相机运动矩阵；再通过相机运动矩阵对图像进行倾斜校正去除相机运动造成的干扰；最后在提取出干净的BVP信号后，通过快速傅里叶变换（fast Fourier transform，FFT）分析幅频特性，利用最大幅值对应的频率（即心率）的原理进行心率检测。为了验证该文算法的鲁棒性，采集了20名受试者的人脸视频进行实验分析，实验结果表明，该文算法优于现有的非接触式心率检测技术，能有效消除相机运动带来的干扰，长期稳定地检测心率。

In view of the problems of inaccurate extraction of blood volume pulse(BVP) wave signal and low detection accuracy caused by motion interference due to camera shaking in the existing video based heart rate detection methods, this paper proposes a camera motion interference resistant heart rate detection algorithm, which uses the image stabilization processing of shaking video to obtain stable facial video for heart rate detection. Firstly, the feature points are selected by the speeded up robust features(SURF) algorithm, the feature matching is carried out, and the camera motion matrix is solved. Then, the image is tilted and corrected by the obtained camera motion matrix to remove the interference caused by camera motion. Finally, the clean BVP wave signal is extracted, the amplitude frequency characteristics are analyzed by fast Fourier transform(FFT), and the heart rate is detected by using the principle that the frequency corresponding to the maximum amplitude is the heart rate. In order to verify the robustness of this algorithm, the facial videos of 20 subjects are collected for experimental analysis. The experimental results show that this method is superior to the existing non-contact heart rate detection technology and can effectively eliminate the interference caused by camera motion. The heart rate can be detected stably for a long time.