Smart Light For Fall Detection Of Elder Living Alone Based On FPGA

The security of the elderly has become a social issue of common concern as China's ageing society.The fall caused great threat to the safety of the empty nest elderly.Because it often causes serious consequences such as loss of consciousness,shock or death if the elder get no relief after fall occurred long time.This thesis uses Filed Programmable Gate Array(FPGA)to realize real-time fall detection with video monitoring and uses the buzzer and household lighting to achieve sound and light alarming.At the same time,notice the community staff and the elderly relatives via General Packet Radio Service(GPRS).The main research work in this thesis is as follows:The design of moving target detection algorithm.This thesis introduces three kinds of common video moving object detection algorithms,and analyzes the principles,advantages and disadvantages in detail.This thesis chooses the frame difference method as the core algorithm to detect moving target because of its simple algorithm,better detection and more easier FPGA implement.The design of moving target fall decision algorithm.First of all,the minimum bounding box is calibrated.Secondly,the fall case can be judged out by the aspect ratio and effective area ratio of bounding box.Sometimes it can not accurately distinguish the daily exercise or real fall such as leg press,push-ups,etc..At the end of the thesis,the judged result is modified based on the change of body centroid.So it can reduce the rate of false positives effectively.The design of FPGA functional modules.Using Verilog Hardware Description Language(Verilog HDL)to configure the FPGA inner logic function.First of all,the image acquisition format will be configured through the Inter-Integrated Circuit(I2C)bus.Then the body target will be extracted through the 4port Synchronous Dynamicrandom access memory(SDRAM)and asynchronous FIFOs.The fall will be determined using the fall decision algorithm.Finally,FPGA drives GPRS module to send falling messages through the Universal Asynchronous Receiver/Transmitter(UART)and drives the sound and light alarming module to alarm through GPIOs if the elder falls.The experimental results demonstrate the frame processing speed is 24.86 fps and the responsing time for alarm is 0.51 s.So this system satisfices the requirement of real-time and makes the elder get timely rescue.Besides,the accuracy of this fall detection system is up to 96%.Its low misforecast rate can reduce the disability rate and mortality greatly.There will be broad development prospects in the field of smart home.