Medical Protection System Based On Intelligent Video Surveillance

There's no doubt that fall is a great threat to the health of the elderly.If there exists surveillance system that can immediately alert and save the real-time images when the fall happens,it will be of great significance and contribution to the timely rescue of the elderly.In this paper,the medical protection system based on intelligent video surveillance aims to detect abnormal fall behaviors promptly,record and remind guardians of immediate ambulance.In response to the above needs,several work has been done as follows.First of all,to satisfy the need of fall monitoring function,a set of intelligent video surveillance medical protection system is designed and implemented.The system consists of camera module,image processing and identification module,alarm module and embedded hardware platform,among which each module has features of clear function,low coupling and flexibility and convenience to expand and maintain.Actually,it has been proved in the test that the system is capable of detecting the target behavior information in time and taking necessary measures in time.Secondly,on account of core position for fall detection module in this system,a fall detection algorithm based on video surveillance is presented.Based on the OpenCV platform,the algorithm first performs a series of preprocessing processes on the data collected by the camera,like noise reduction and grayscale,then detects the moving targets by using mixed Gaussian background modeling.Then,aiming at overcoming the shortcomings of high computational complexity and poor anti-interference of background modeling,an improved hybrid Gaussian background modeling algorithm is proposed.By adaptively selection of the number of Gaussian bases and the adjusted learning rate,the efficiency and recognition rate of traditional algorithms can be optimized.According to the characteristics of fall behavior,three-level classification detector the filter the foreground image fall identification.Experimental results show that the algorithm can precise detection of fall behavior.Finally,all the requisite algorithm is transplanted to the Raspberry Pi platform and a set of software is designed to capture video information processing and sharing through Mjpg-streamer video server equipped on the Raspberry Pi platform.In addition,it enables external access to monitor video information through the mapping port Ngrok tools.It's indicated in the test process that the entire system is able to achieve a real-time identification,monitoring and protection.