Non-Contact Blood Pressure Measurement Based on IPPG

Blood pressure is an important physiological parameter to reflect human vital signs.In order to achieve the non-contact dynamic blood pressure acquisition based on ordinary optical camera,a theoretical understanding of the functional relationship between blood pressure and pulse wave signal conduction time.And through imaging photoelectric plethysmography(IPPG),pulse wave signal conduction time of forehead and hand was obtained with ordinary optical camera.First,the pulse wave conduction time was obtained by recording the video with an ordinary optical camera.Second,real-time blood pressure values were collected.Finally,based on the relationship between blood pressure and pulse wave conduction time,a non-contact blood pressure measurement prediction model was obtained through neural network fitting.So that non-contact blood pressure measurement with optical camera could be completed.The method in this paper has several advantages,such as low requirements on measuring equipment,low cost,and simple operation.It can let people get rid of the discomfort caused by measuring equipment such as cuff and can measure blood pressure at any time.The predicted blood pressure results were compared with an Omron wrist electronic sphygmomanometer.The calculated error of systolic blood pressure is-9.28%~3.16%,and the error of diastolic blood pressure is-9.84~4.35%.

Elderly Fall Detection Based on Improved SSD Algorithm

We propose an improved a single-shot detector(SSD)algorithm to detect falls of the elderly.The VGG16 network part of the SSD network is replaced with the MobilenetV2 network.At the same time,we change the infrastructure of MobilenetV2 network,the three layers that were not down-sampled at the end were removed,which can make the model structure simpler and faster to detect.The complete Intersection-over-Union(CIoU)loss function is introduced to get a good regression of the target borders that have different sizes and different proportions.We use Feature Pyramid Network(FPN)for up-sampling,it can fuse low-level feature maps with high resolution and high-level feature maps with rich semantic information.For sampling results,we use the Secure Shell(SSH)module to extract different receptive fields,which improves the detection accuracy.Our model ensures that the accuracy of the elderly fall detection remains unchanged,but it greatly improves the detection speed that only takes 10 milliseconds to detect a picture.