The Design of Binocular Human Fall Detection System Based on Fusion of Vision and Multi-sensor Algorithm

According to WHO statistics, fall death is the second leading cause of accidental death, among which the elderly have the highest risk of death or serious injury due to falls. Real-time detection and accurate medical treatment after a fall can effectively reduce the risk of death. At present, the implementation of detection systems for detecting human fall is mainly based on the following three categories: wearable devices, monitoring systems based on environmental sensors, and visual monitoring. Wearable devices require contact with the body, making them difficult to operate and not highly acceptable; The monitoring systems based on environmental sensors are mainly infrared radar and millimeter wave radar, and the accuracy of fall detection is currently relatively low; Visual monitoring is currently mainly based on monocular cameras, which cannot obtain depth information and has low accuracy. This system combines the advantages of machine vision and multiple sensors, and has been upgraded from a monocular camera to a binocular camera, which can more accurately obtain depth information and improve fall detection accuracy.