基于深度图像的非接触式呼吸检测算法研究

呼吸是人的基本生命活动，监测呼吸可以得知呼吸道和胸廓运动的生理、病理学状态，对某些呼吸系统疾病的诊断有重要的参考价值；提出了一种非接触式呼吸监测方法：对红外视频流中的每帧胸腹部区域数据进行降维，计算所有胸腹部区域数据的方差，将一定时间段内的方差序列进行低通滤波。最后根据方差序列可以获得该段时间内的呼吸频率和呼吸暂停时间；提出的非接触式呼吸检测算法在不影响被监测者正常睡眠活动的情况下，可以准确获取呼吸频率与其他相关参数，为健康监测和相关疾病的诊断提供了数据支持；日常家居场景的实验中，检测到的呼吸次数与实际完全一致，并且与实际胸腹部起伏变化基本同步，较好的保证了结果的准确性。

Research on a Non-contact Breathing Detection Algorithm Based on Depth

Breathing monitoring plays an important role in monitoring respiratory physiology of chest movement, pathological condition, and the diagnosis of certain diseases of the respiratory. Presents a method for non-contact monitoring of breathing. Dimensionality reduction is applied to the infrared region of the video stream data chest and abdomen. The standard deviation of the data in the chest and abdomen areas is then calculated within a time window. A sequence of standard deviations are generated and passed to a low-pass filter. The proposed non-contact monitoring of breathing method can get an accurate respiratory rate and other parameters for the monitoring and diagnosis of health related diseases without affecting normal sleep activity. Respiratory rate and breathing pause time can be obtained based on the filtered standard deviation data.We carried out experiments in a home-based environment. The experimental results show that our developed system can calculate the number of breaths that is consistent with the actual number of breaths.