Non-contact Measurement Of The Human Respiratory Rate

The continuous development of society has improved people’s awareness of health. Recently there is increasing interest in ubiquitous healthcare that provides diagnosis through the real-time dynamic monitoring method to monitor the human health. Respiratory monitoring is not only an important aspect of human health, but also an important part of monitoring instruments in hospital and wearable devices. But the traditional methods of measurement of respiration usually get the respiratory signals by a variety of sensors, which make them inconvenient, and the price is relatively expensive. It is very difficult to be popular with more people. So this paper discusses the method of respiration measuring in non-contact conditions.In this paper, we obtained the sampled signals from camera color video, which includes human face and chest; then used a variety of image processing algorithms for image processing to extract human face and chest area, as the signal acquisition area. Then we extracted the regional gray value signal, processed the data to get more accurate respiratory wave by efficient signal-denoising method, and on this basis, calculate the respiratory rate, so as to realize the real-time monitoring of respiration. On the other hand, this paper also presents certain exploration to extract physiological signal from a complete face video.The main research contents are as follows:(1) Detecting face based on skin color. The object of this paper to detect is the chest. But there is no obvious features on the chest area. This paper detected the face at the beginning, then calculated the chest position. We obtained the face skin color area by the illumination compensation, image filter, skin color segmentation, morphological processing of digital image processing technology. We used a projection method to obtain the boundaries of face and chest area.(2) Extracting the respiratory waveform. We got the gray values of the chest area in each frame, analyzed the signal spectrum and the possible presence of noise, used wavelet analysis to filter out the noise of human jog, ambient light interference, etc., to obtain relatively clean breathing signals.(3) Designing the experiments includes three aspects, accuracy, dynamic, and practicality.designing the preliminary interface.(4) The extraction of heart and breath signal from the whole video. This paper also introduces a new method, using color facial video, and obtaining the useful signals based on face detection and blind source separation of color channels. Empirical mode decomposition(EMD) is an adaptive method, commonly used in the analysis of non-stationary signals. In this paper, EMD method was used to decompose the signal into intrinsic mode functions(IMFs) that can reflect the life of information, then according to the criterion of extraction, extracted accurate heartbeat and respiration signals. Bland-Altman method was used to analyze the experimental data.This paper used ordinary camera system, developed software systems and related algorithms to measure human respiration in real-time. By collecting the human front video, we obtained physiological parameters of the measured person which can provide support for the convenience of subsequent research health care. Experimental results show that the method can measure real-time and long-term human breathing quickly and accurately measure, real-time and with better accuracy.