Research On Respiratory Rate Detection Method Via Video Processing

Respiratory rate is one of the important indicators for evaluating the health sta tus of human body and it has great significance in the prevention and diagnosis of respiratory diseases.At present,respiratory detection equipment is mostly hos-petal-specific equipment that cannot meet the needs of daily household testing.T-he non-contact video-based respiratory detection method has great application pro-sects in the field of smart medical care because of its versatility and low cost.However,the existing video-based respiratory ’detection methods still have some problems in scene noise reduction and night signal acquisition,and the acquisitio n-on of effective respiratory area under different detection scenarios affects the re spiratory detection performance.Given that the problems for noise signal suppression and scene applicability,t-his paper deeply studies the method of video magnification based on phase infor mation and the region tracking method of respiratory characteristics,proposes a method for respiratory rate detection based on chest positioning in day and night environments and respiratory rate detection method based on facial features.The innovations are summarized as follows:A novel,video-based respiratory rate detection method based on chest location is proposed to study the effective extraction of signals in the day and night envi ronment.Firstly,an infrared camera was used to ensure the effective acquisition of night respiratory signals,and face detection and the proportion of human body determined the chest region;considering the changes of respiratory phase inform-ation,we used phase motion magnification technology to achieve motion signal magnification to extract the respiratory signal effectively.We applied the maxim-um likelihood criterion to estimate the first respiratory rate,then we used that re spiratory rate as a reference to process the original wave by smoothed filtering.Finally,accurate respiratory rate was estimated by peaking detection.A novel video-based respiratory rate detection method based on facial feature i nformation is prosed to study the breathing signal extraction in facial region.Aft er obtaining the face area,we use DRMF(discriminative response map fitting)to generate feature points of facial organs.Then we chose the breathing features as tracking target,and recorded the change of the longitudinal displacement of the f eature point as a respiratory trajectory.Introducing the wavelet decomposition to eliminate baseline drift noise of tracking data,we could obtain stable respiratory signal.In addition,a new method based on IPPG signal is proposed,we verified the feasibility and scope of the method finally.To verify the performance of the non-contact video-based respiratory rate detec tion method proposed in this paper,we collected videos under day and night con ditions and conducted a quantitative analysis of the result of respiratory detection The results showed that in experiments with indoor light sources,the propose-d method has good consistency with the actual respiration rate.Compared with the existing video-based respiration detection method,our method has higher accurac-y and better robustness.In addition,respiratory rate detection based on facial fea ture information has reliable detection performance,which provides a feasible s-o lution for realizing respiratory rate detection in mobile phones or other mobile te rminals in the future.