提高基于机器视觉的心率测量精度的方法（英文）

发布时间：2018-09-18 13:36

【摘要】：基于摄像头(可见光)的心率测量方法,能够以非接触的方式检测受试者心率,无论在临床应用还是在家庭健康监护中都有着极大的应用前景。但是,CMOS摄像头的行扫描方式采集图像以及计算机系统的图像采集时钟抖动都会影响此类心率测量的精度,从而引入相位误差和系统随机误差。本文针对这两大主要误差进行研究分析,提出了消除相位误差的基于傅里叶变换的幅频叠加算法、消除图像采集系统时钟抖动误差的基于时间标定的三次样条插值重构方法。幅频叠加算法只在幅频域对信号进行处理,能够忽略信号的相位影响,从而消除相位误差。时间标定的三次样条重构算法能够重构图像的均匀采集,从而消除系统时钟抖动引起的随机误差。同时,通过理论推导论证了两种算法的可行性,进而利用LED模拟实验和实际心率测量实验验证该算法在提高信号检测精密度中的特性。在模拟实验中,对每帧图像的200行作幅频叠加运算,能够使信号幅值相对提升4.58%;基于时间标定的三次样条重构算法能使检测的信号频率的均方根误差缩小30%以上;在实际心率检测中,幅频叠加算法可使心率幅值相对提升达33.5%,基于时间标定的三次样条重构算法可使心率准确度提升40%左右。因此,模拟实验和实际心率测量实验均验证了提出的算法在提高系统检测精度上的有效性,提高了基于机器视觉心率测量的抗干扰能力。这两种算法不仅能够提高基于机器视觉的心率检测精度,而且能够适用于基于机器视觉对一定频率信号的检测,在机器视觉检测中具有广泛的意义。
[Abstract]:The method of heart rate measurement based on camera (visible light) can detect the heart rate of subjects in a non-contact manner, which has great application prospects in both clinical and family health monitoring. However, the line scanning mode of CMOS camera and the clock jitter of image acquisition in computer system will affect the accuracy of this kind of heart rate measurement, thus introducing phase error and system random error. In this paper, the two main errors are studied and analyzed, and a Fourier transform based amplitude-frequency superposition algorithm is proposed to eliminate the phase error, and a cubic spline interpolation reconstruction method based on time calibration is proposed to eliminate the clock jitter error of the image acquisition system. The amplitude-frequency superposition algorithm only processes the signal in the amplitude-frequency domain, which can ignore the phase effect of the signal and eliminate the phase error. The cubic spline reconstruction algorithm based on time calibration can reconstruct the uniform image acquisition and eliminate the random error caused by the system clock jitter. At the same time, the feasibility of the two algorithms is proved by theoretical derivation, and the LED simulation experiment and the actual heart rate measurement experiment are used to verify the characteristics of the algorithm in improving the precision of signal detection. In the simulation experiment, 200 lines of amplitude-frequency superposition operation on each frame image can make the signal amplitude increase by 4.58, and the cubic spline reconstruction algorithm based on time calibration can reduce the root mean square error of the detected signal frequency by more than 30%. In the actual heart rate detection, the amplitude-frequency superposition algorithm can increase the amplitude of heart rate to 33.5, and the cubic spline reconstruction algorithm based on time calibration can improve the accuracy of heart rate by about 40%. Therefore, the simulation experiment and the actual heart rate measurement experiment both verify the effectiveness of the proposed algorithm in improving the detection accuracy of the system, and improve the anti-interference ability of heart rate measurement based on machine vision. These two algorithms can not only improve the accuracy of heart rate detection based on machine vision, but also can be used to detect certain frequency signals based on machine vision.
【作者单位】：天津大学精密测试技术及仪器国家重点实验室;天津大学天津市生物医学检测技术与仪器重点实验室;天津中医药大学中医药工程学院;
【基金】：National Natural Science Foundation of China(NSFC)(81202643) Tianjin Application Basis&Front Technology Study Programs(14JCZDJC33100)
【分类号】：TP391.41

【相似文献】