基于面部视频的血流信号提取方法的比较及适用性研究

发布时间：2019-06-15 17:58

【摘要】：基于独立分量分析(ICA)的盲源分离技术可从面部视频序列中分离出血流脉冲信号(BVP),进而可获取心率、血氧饱和度和呼吸率等生命体征参数。但现有研究结果并未有效展示ICA在BVP获取方面的优势,因此,一些研究者建议直接利用面部视频图像序列的绿色通道分量(G分量)并结合时域滤波进行BVP提取(G-BVP)。本文基于9位受试者所提供的面部视频数据,对ICA-BVP和G-BVP两种方法的性能和特点进行了比较,研究了面部器官自然运动和光照变化等因素引起的视频干扰成分对BVP提取效果的影响。实验表明,ICABVP方法在消除运动和光照干扰方面具有较明显的优势。在ICA-BVP方法的具体实现过程中,提出了基于谱峭度的BVP信号自动识别方法,用以解决ICA输出排序不确定的问题。该方法在动态BVP信号获取和心率估计实验中取得了较理想的测试效果。综上所述,ICA-BVP方法在基于面部视频的血流信号提取及生命体征参数估计的应用中,具有更好的稳定性和准确性。
[Abstract]:The blind source separation technique based on independent component analysis (ICA) can separate the bleeding flow pulse signal (BVP), from the facial video sequence, and then obtain the vital sign parameters such as heart rate, blood oxygen saturation and respiratory rate. However, the existing research results do not effectively show the advantages of ICA in BVP acquisition. Therefore, some researchers suggest using the green channel component (G component) of facial video image sequence directly and combining with time domain filtering for BVP extraction (G-BVP). Based on the facial video data provided by 9 subjects, the performance and characteristics of ICA-BVP and G-BVP methods were compared, and the effects of video interference components caused by natural movement of facial organs and light changes on BVP extraction were studied. The experimental results show that the ICABVP method has obvious advantages in eliminating motion and light interference. In the implementation of ICA-BVP method, an automatic recognition method of BVP signal based on spectral kurtosis is proposed to solve the problem of uncertain ICA output sequencing. This method has achieved ideal test results in dynamic BVP signal acquisition and heart rate estimation experiments. To sum up, ICA-BVP method has better stability and accuracy in the application of blood flow signal extraction and vital sign parameter estimation based on facial video.
【作者单位】：安徽大学计算机科学与技术学院;安徽大学计算机智能与信号处理教育部重点实验室;
【基金】：国家自然科学基金(61271352,61401002)
【分类号】：TP391.41

【相似文献】