Video Respiratory Rate Detection Algorithm Based On Motion Feature

Respiratory rate is an important Physiological indicator of human,which plays an irreplaceable role in evaluating human respiratory status and diagnosing respiratory diseases.At present,the breathing rate detection method requires the operation of professionals,and there are still shortcomings in the daily detection of the family.In recent years,non-contact respiratory rate detection methods have the advantages of low cost and strong applicability,which has great potential for development in the field of smart medicine.The existing video-based respiratory rate detection methods are still insufficient in real time and have limitations on human posture,which ultimately affects the detection performance.To solve this problem,this paper further studies the video respiratory rate detection algorithm based on motion features,which is summarized as follows:(1)A rapid respiratory rate detection method based on motion feature prediction was proposed.Through the estimation of the breathing movement area and direction in the video,the phase-based rapid detection of respiratory rate is realized.First,the chest area was selected by face detection and head-to-body ratio.Then the respiratory signal model was established,the precise breathing area was obtained by the maximum likelihood method,and the main movement direction in the respiratory area was determined by the gradient information.Finally,combined with the direction of respiratory movement,the phase based signal processing algorithm was used to extract respiratory signals and obtain respiratory rate.(2)A multi-pose respiratory rate detection method based on the consistency of respiratory movement is proposed,which can detect the respiratory region without face information.Specifically,the algorithm first established the saliency model of the fusion of respiratory features,and the respiratory region in the model had higher significance.Then,according to the saliency model,the respiratory area was obtained.Finally,the respiratory rate was obtained.The experimental results show that the algorithm can detect respiratory rate in multiple postures of human body.In order to verify the performance of the video respiration rate detection algorithm based on motion features proposed in this paper,a large number of human respiration videos under different conditions were collected for experiments.The experimental results showed that under normal sitting posture,the time of detecting respiratory rate by the fast method based on motion feature prediction was significantly shortened,and it was in good agreement with the real value.In other postures,the respiratory rate detection method based on the consistency of breathing movement can accurately select the breathing area and obtain the accurate respiratory rate.