基于FastICA算法的非接触心率、呼吸频率检测

发布时间：2017-12-31 20:30

本文关键词：基于FastICA算法的非接触心率、呼吸频率检测　出处：《南昌大学》2015年硕士论文　论文类型：学位论文

【摘要】：近几年来我国很多城市,尤其是北京,饱受雾霾的侵扰,雾霾对人体的呼吸系统产生了极大的危害,人们呼吸道疾病发病率逐年升高,心血管方面疾病发病率也有所增加。心率作为心血管疾病检查的最基础表征指数,对人体的健康指标有重要意义。快速和方便的心率测量成为日常生活的一个方面。本课题基于视频图像处理,将一种快速独立分量分析的方法(FastICA)运用到人脸图像信号处理领域,并由此测量出心率。在自然环境下,通过网络摄像头采集人脸彩色视频,从视频流中检测出人脸并提取心率信号,实现简单的非接触式心率测量。本课题的研究对人们便捷检查自身心血管方面的健康,降低未知的危险,有着十分重要的意义。该方法在不接触人体皮肤的情况下,间接检测出人的心率。相比较传统的接触式心率检测的方法,这种检测方法具有无创、简单、高效等特点。基于光电容积描记法(PPG)的原理,本文先综述了基于人脸视频的非接触心率检测的研究历史、现状和理论基础,然后将基于FastICA算法的独立分量分析运用到观测信号的盲源分离。该算法能对混合信号进行某种最优线性分解并获取统计独立的分量。从三个原始的观测信号分解得到三个独立信号,对信号进行相关性分析、平滑、滤波、插值等得出信号周期,间接计算出心率。人的脸部图像信息还包含了另一项生理参数——呼吸频率。因为信号的高频分量和呼吸频率是有联系的,本文对最终得到的心博周期信号采用频谱分析,并从功率谱中估计呼吸频率。最后,通过实验仿真验证,对比手指式血氧饱和仪的测量数据,本文验证了该方法的可靠性。
[Abstract]:In recent years many of our city, especially in Beijing, suffering from haze intrusion, haze on the human respiratory system brought tremendous harm to people, incidence of respiratory diseases increased year by year, the incidence of cardiovascular disease also increased. The most basic characterization index of heart rate as a cardiovascular disease examination, are important health indicators on the human body. The rapid and convenient measurement of heart rate has become one aspect of daily life. This paper based on video image processing method, a fast independent component analysis (FastICA) is applied to face image signal processing field, and thus the measured heart rate. In the natural environment, through the network camera face color video to detect the face, and the extraction rate of signal from the video stream, non-contact measurement of heart rate is simple. The research of the people convenient to check their cardiovascular health Kang, reduce the unknown risk, has very important significance. This method is not in contact with the human skin under the condition of a person's heart rate. The indirect detection method for contact detection rate compared with the traditional, this method is non-invasive, simple, efficient and so on. Based on the method of photoplethysmography (PPG the principle), this thesis first reviews the research history of non contact rate of face detection based on video, status and theoretical basis, and then the independent component analysis FastICA algorithm applied to blind source separation based on the observed signal. This algorithm can of mixed signal in some optimal linear decomposition and obtain statistically independent components from three. The original signal is decomposed into three independent observation signal, correlation analysis for signal smoothing, filtering, interpolation and so that the signal cycle, and calculate the heart rate. The human face image information also contains another student Physical parameters of respiratory frequency. Because of the high frequency component and respiratory frequency signals are connected, the cardiac cycle signal obtained by spectrum analysis, and estimate the respiratory rate from the power spectrum. Finally, through simulation experiments, comparing with the measurement data of finger type blood oxygen saturation instrument, this paper verifies the reliability of the method.

【学位授予单位】：南昌大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：R443;TP391.41

【共引文献】