Posture Recognition Based Fall Detection System

Posture Recognition Based Fall Detection System A POSTURE RECOGNITION BASED FALL DETECTION SYSTEM FOR MONITORING AN ELDERLY PERSON IN A SMART HOME ENVIRONMENT ABSTRACT The mobile application is capable of detecting possible falls for elderly, through the use of special sensors. The alert messages contain useful information about the people in danger, such as his/her geo location and also corresponding directions on a map. In occasions of false alerts, the supervised person is given the ability to estimate the value of importance of a possible alert and to stop it before being transmitted. This paper describes system for monitoring and fall detection of ELDERLY PEOPLE using triaxial accelerometer together with ZigBee transceiver to detect fall of ELDERLY PEOPLE. The system is composed of data acquisition, fall detection and database for analysis. Triaxial accelerometer is used for human position tracking and fall detection. The system is capable of monitoring ELDERLY PEOPLE in real time and on the basis of results another important parameters of patient can be deducted: the quality of therapy, the time spent on different activities, the joint moveme nt, etc. The system, including calibration of accelerometers and measurement is explained in detail. The Accidental Fall Detection System will be able to assist carriers as well as the elderly, as the carriers will be notified immediately to the intended person. This fall detection system is designed to detect the accidental fall of the elderly and alert the carriers or their loved ones via Smart-Messaging Services (SMS) immediately. This fall detection is created using microcontroller technology as the heart of the system, the accelerometer as to detect the sudden movement or fall and the Global System for Mobile (GSM) modem, to send out SMS to the receiver. INTRODUCTION The leading health problems in the elderly community. They can occur in home as well as in hospitals or in the long-term care institutions [1]. Falls increase risk for serious injuries, chronic pain, long-term disability, and loss of independence, psychological and social limitations due to institutionalization. Nearly 50% of older adults hospitalized for fall- related injuries are discharged to nursing homes or long-term care facilities [2]. A fall can cause psychological damage even if the person did not suffer a physical injury. Those who fall often experience decrease activities of daily living and self-care due to fear of falling again. This behavior decreases their mobility, balance and fitness and leads to reduced social interactions and increased depression. The mortality rate for falls increases progressively with age. Falls caused 57% of deaths due to injuries among females and 36% of deaths among males, age 65 and older [3]. Majority of falls result from an interaction between multiple long-term and short-term factors in person’s environment [4]. Common risk factors include problems with balance and stability, arthritis, muscle weakness, multiple medications therapy, depressive symptoms, cardiac disorders, stroke, impairment in cognition and vision Detection of a fall possibly leading to injury in timely manner is crucial for providing adequate medical response and care. Present fall detection systems can be categorized [7, 8, 9] under one of the following groups: User activated alarm systems (wireless tags), Floor vibration-based fall detection, Wearable sensors (contact sensors and switches, sensors for heart rate and temperature, accelerometers and gyroscopes ), Acoustic fall detection, Visual fall detection. The most common method for fall detection is using a triaxial accelerometers or bi-axial gyroscopes. Accelerometer is a device for measuring acceleration, but is also used to detect free fall and shock, movement, speed and vibration. Using the threshold algorithms while measuring changes in acceleration in each direction, it is possible do detect falls with very high accuracy. Using two or more tri-axial accelerometers and combining them with gyroscopes at different body locations it is possible to recognize several kinds of postures (sitting, standing, etc.) and movements, thereby detecting falls with much better accuracy. An easy and simple method to detect fall detection of ELDERLY PEOPLE is using accelerometer together with ZigBee transceiver to communicate with Monitoring System through wireless network, and in this paper a system for monitoring and fall detection of ELDERLY PEOPLE using mobile MEMS accelerometers will be presented. . The first three functions provide recording in a database, and also a text message is sent to the supervisor with latitude, longitude and other useful data. Afterwards, you can detect the elder person through Google maps. Additionally, an application was implemented for the attending physician, which is connected with the database, through which s/he can obtain a complete picture of the patients’ status, to draw useful conclusions and proceed to possible change in medical treatment. EXISTING SYSTEM: An application for Apple IOS by using an accelerometer to detect falls. A possible drawback is that the development platform Apple IOS is not accessible to the average user. An application in Symbian s60 using machine learning algorithm takes 64 samples every two seconds from the accelerometer and decides whether there is a fall. PROPOSED SYSTEM: In this paper, we designed an application with the ability of automatic fall detection, by using the mobile sensors, warning signal by pressing a button in cases of emergency, detection and automatic notification to supervisors as well as visual display to passerbies. The application uses basically two incorporated mobile sensors, namely the accelerometer and the gyroscope sensor.A counter starts counting loudly on the screen from 30 to 0. If the counter reaches 0, then an SMS message is sent to the caregiver or relative and an entry is made to the Database. The first service detects the patient’s position and calculates whether the patient is further away than a set distance. When activated can give directions to the patient what route to follow to return back to home. APPLICATIONS: Automatic fall detection. Warning if the elder moves away from the place of residence directions given on the map. ADVANTAGES: Elder’s safety can be assured. Fast First aid or medical treatment can be guaranteed. DISADVANTAGES: Device Sensor should be carried out whenever the person moves over. SYSTEM DESIGN Architecture Diagram SYSTEM FOR MONITORING AND FALL DETECTION The whole system consists of a set of sensors (two or more sensors on the patient, usually MEMS sensors) which the patient wears on himself, local units to collect data that are placed in patient vicinity and systems for collecting. The tiny sensors in the strap are capable of measuring user orientation and motion in three-dimensions and it is constantly monitoring and analyzing the signals in real-time looking for movement indicating a fall. From the comparison Table Error! No text of specified style in document. .1, it shows that the system maybe a hindrance to the consumer in terms of price over the years. The aim of this project is to be able to provide equal standard of care at an affordable cost. The system is shown in Figure 1 the space is divided into sections which are defined by interior and exterior of the institution in which a system is operated. Each room is stocked with local receivers. Local receivers collect data from sensors that the ELDERLY PEOPLE are wearing on the clothes. The sensors are small and lightweight. One sensor is located in the upper garment and the other at the bottom. This is not limited to two sensors, if necessary, there may be more, but for the detection of falls to the back the system must have at least 2 sensors Local receivers pass information to the server. The server information is processed local health care service. Personal computers are used to browse the database collected a t the server. The database contains information about the mobility of ELDERLY PEOPLE, treatment efficacy, joints. All these data can be analyzed offline and used to adjust patient therapy. This has served a double function of the system Real-time patient monitoring and early detection of the fall in order to deliver medical assistance as soon as possible. In this application Free scale TM ZSTAR wireless sensing triple axis board is used (Fig. 2). It is very practical because of low power consumption, portability, and the ability to be mounted in small pockets inside the clothes of ELDERLY PEOPLE. Board is divided into sensory and receiver part. The sensor is placed at the patient and is equipped with an accelerometer, microprocessor, and transceiver with the antenna which sends the measurement data to the receiver. The receiver also has a microprocessor that adjusts the signals received through the antenna to send with the USB protocol. These data are sent to the server. The server collects process and stores the data. Each sensor that is connected to the patient is personalized, and its data are stored in a file under persons name to get an overview of all activities and physical stress of the patient FALL DETECTION USING TWO ACCELEROMETERS In this chapter the operation of the system through one of its functions and to the detection of fall will be described. The figures have been simplified for better understanding of the system. The algorithm used is improved algorithm given in, with better detection of backwards falls. Setup for accelerometer fall detection, consists of the measuring sensors with transmitter, receiver and server for data processing and fall detection. The fall is detected by the algorithm described in. It can be seen that fall detection algorithm uses data from both sensors that are monitored at the same time. This algorithm is able to distinguish between falls (forward, back word fall into a sitting position) and the normal daily activity, such as walking, mastering stairs, sitting in a chair, lying walking is also detecting by the sensors. However, these impacts are not isolated, and after them there is no significant change in orientation between the two sensors. Vectors are in the area that will call common zone .if an isolated stoke which causes a change in orientation of the body is detected, or the orientation of certain body parts in relation to the situation before the stroke, then with some certainty it can be said that the fall had occurred. Dataflow Diagram SYSTEM IMPLEMENTATION Modules details Phase 1 Modules Fall Detection Location Tracking Phase 2 Modules Communication Route Map Integration Fall Detection: The FALL DETECTION is something that we have developed at Alert1 so you can be safe at all times. Whether you are a senior citizen and want to maintain your independence, a concerned family member looking for peace of mind, or a caregiver with patients, this tool has been developed for you. Prevention is key. Use it to inspect and detect hazardous areas in your home that could result in a fall. If you answer no to the questions, you have already taken action to reduce your risk of falling. If you answer yes to any of the questions, consider making the recommended change or adaptation to reduce your risk of falling. Location Tracking: Real-time locating systems (RTLS)are used to automatically identify and track the location of objects or people in real time, usually within a building or other contained area. Wireless RTLS tags are attached to objects or worn by people, and in most RTLS, fixed reference points receive wireless signals from tags to determine their location. The physical layer of RTLS technology is usually some form ofradio frequency(RF) communication, but some systems use optical (usuallyinfrared) or acoustic (usuallyultrasound) technology instead of or in addition to RF. Tags and fixed reference points can be transmitters, receivers, or both, resulting in numerous possible technology combinations. RTLS are a form oflocal positioning system, and do not usually refer to GPS,mobile phone tracking, or systems that use only passiveRFIDtracking. Location information usually does not include speed, direction, or spatial orientation. Communication: The table that maintained the mapping between the agent’s name and the landmark location is shared and updated by the agents who were on nodes within the landmark’s coverage. When the node is not a landmark node, the table is used as a cache table. If communication with the other agent succeeds, the locations and the agent names are registered in this cache table. It is possible for the agent to periodically get the location of the target agent and store it in the cache table. The use of a cache table enables agents to initiate direct communication with each other and reduce the communication overhead to landmarks. When the cache misses, the agent sends a request to the landmark to get updated information. Agents can also delete the information from the cache table. The communication between landmarks is implemented, however we only use this communication to call the target agent when there is no target agent within the coverage area. This primitive is used when the pro grammer deploys agents and makes deployment of agents easy. Routemap Integration: The integration of spatial maps in mobile was investigated using a spatial analog to sensory preconditioning. The GPS chip outputs the positioning information which is transferred over a GPRS link to the mobile operator’s GGSN (Gateway GPRS Support Node) and then to a remote server over a TCP connection. The TCP server stores the incoming positional data in a mySQL database. When a user clicks on the tracking page., Zope, which is an open source web application server, serves up an HTML page with an embedded javascript code. The javascript would run in the users browser and has instructions to retrieve the positional information from the mySQL database every second. It then integrates this information into Google Maps through Google Maps API which displays the position on a map. Since the positional information is retrieved every second and the maps updated at the same frequency, a real time GPS tracking effect is achieved. CONCLUSION Triaxial accelerometers can be used for detecting fall of ELDERLY PEOPLE. They offer low cost solution, and together with wireless connectivity solutions such as ZigBee provide efficient solution for both ELDERLY PEOPLE and medical personnel l. In this paper I have presented an intelligent mobile multimedia application that can be incorporated into modern mobile smartphones in order to be used for the needs of the elderly. It is in our future plans to evaluate this system in order to test its efficiency in actually helping these people sufficiently. It is also in our future plans to extend the system’s capabilities by incorporating new services. These services include the following: Embed a belt measuring heart rate as an external sensor Integrate a gyroscope sensor instead of an orientation sensor, for more accurate results Integration of social networks to alert senders Integrate public agency to alert senders Add a system administrator feature. References: A. Chan and N. Vasconcelos, â€Å"Counting people with low-level features and Bayesian regression,† IEEE Trans. Image Process., vol. 21, no. 4, pp. 2160–2177, Apr. 2012. E.Auvinet, F. Multon, A. Saint-Arnaud, J. Rousseau, and J. Meunier, â€Å"Fall detection with multiple cameras: An occlusion-resistant method based on 3-d silhouette vertical distribution,† IEEE Trans. Inf. Technol. Biomed., vol. 15, no. 2, pp. 290–300, Mar. 2011. Y. Hou and G. Pang, â€Å"People counting and human detection in a challenging situation,† IEEE Trans. Syst. Man, Cybern. Part A: Syst. Humans, vol. 41, no. 1, pp. 24–33, Jan. 2011 Y. Chen, L. Zhu, A. Yuille, and H. Zhang, â€Å"Unsupervised learning of probabilistic object models (POMs) for object classification, segmentation, and recognition using knowledge propagation,† IEEE Trans. PatternAnal. Mach. Intell., vol. 31, no. 10, pp. 1747–1761, Oct. 2009 F. Lecumberry, A. Pardo, and G. Sapiro, â€Å"Simultaneous object classification and segmentation with high-order multiple shape models,† IEEETrans. Image Process., vol. 19, no. 3, pp. 625–635, Mar. 2010