Designand Implementation Of The Elderly Monitoring System Based On Posture Recognition

With the development of technology and the implement of Family planning, andFamily planning is a basic state policy for China, The population of elderly people hasreached12%, which is pushing china to an aging society. As a result, health careservice manpower is urgent needed. However, human resource is more and moreinsufficient. For the elderly health improvement, health assessment and health care,some innovative methods combined with technical assistance are needed.This paper is devoted to recording and assessing user’s posture everyday withaccelerometer sensors. System proposed in this paper also presents the fall monitoring.The entire system consisted of4modules: main control module, sensor module,Android mobile phone module and the posture recognition module on PC. Maincontrol unit, as known as embedded system, initializes and drives all the sensors andBluetooth, the angle calculation is also finished on embedded system when fallhappens. What’s more, embedded system sends instructions to Android phone whenfall happens. Android phone stores the data from sensors, and sends warning messageto a designated number when fall happens. Sensor module provides raw data ofdifferent postures. This paper mainly finishes these works:First of all, the hardware platform is developed: including chip selection,initialization and driving sensors, communicating between Android phone andembedded system by Bluetooth, and installing sensors on body’s9specific locations.Secondly, a new method for fall monitoring suitable for implementing on MCU ispresented in this paper and the experimental results demonstrate that this methodworks well. Besides, by the advantage of multiple sensors rather than single one, theposture after fall happens can be recognized in order to get a better balance betweenmissing and misinterpretation.Thirdly, a new algorithm for angle calculation suitable for implementation onMCU is designed in this paper. CORDIC algorithm can implement the anglecalculation in two dimensions by some basic adders and binary shifters. After studyCORDIC algorithm carefully, an angle in3dimensions can be calculated byimplementing CORDIC algorithm2times. After many tests on logic analyzer, theproposed algorithm has a very good experimental result in terms of computationalaccuracy and computation time. Lastly, a new posture recognition system working on computer is designed in thispaper. The mentioned system is developed on2platforms: vs2008and Matlab. Firstof all,4basic postures: standing, sitting, lying and walking, are recognized by usingRBF neural networks on Matlab. After that, all the other recognition works arefinished on vs2008, such as the calculation of steps, the balance of two legs, recordingand monitoring of waist posture and other works.