WO2012119903A1

WO2012119903A1 - Method and system for detecting a fall and issuing an alarm

Info

Publication number: WO2012119903A1
Application number: PCT/EP2012/053443
Authority: WO
Inventors: Jörg Hofmann; Jens LÜDERS; Uwe BÖHNKE; Jens Wagner
Original assignee: Deutsche Telekom Ag
Priority date: 2011-03-04
Filing date: 2012-02-29
Publication date: 2012-09-13
Also published as: EP2681722B1; DK2681722T3; EP2681722A1

Abstract

A method and a device for detecting a particular emergency situation or detecting a fall and issuing an alarm is made available. The method comprises monitoring a space by means of at least one optical sensor, such as a video camera or infrared camera, detecting a fall of a person in the monitored space by evaluating the images recorded with the optical sensor, distorting the image data and triggering an alarm and transmitting the distorted image data to at least one authorized receiver if a fall of a person in the monitored space is detected.

Description

Method and system for fall detection and transmission of an alarm

Technical area

The claimed invention relates to a method and system for fall detection, detection of particular emergencies and transmission of an alarm in the home environment. More particularly, the invention relates to a method and system for fall detection by means of a passive optical sensor and passing a multi-level alarm in the form of alienated image data.

State of the art

Losses pose a significant risk of accidents especially for the elderly. These individuals often injure themselves in such a way that they are no longer able to stand up or request help from others, which can lead to a life-threatening situation. Preventive measures to prevent falls can be taken, but falls of just older, single persons can never be completely ruled out, so that a fall detection with connected assistance can increase the quality of life of the people concerned by the fact that he continues to live in the familiar environment and yet a monitoring of the group of people by medical personnel and / or relatives can be done.

US Pat. No. 7,248,172 B2 has proposed a method and system for fall detection, which includes that the person to be monitored carries a device for detecting falls or for triggering an alarm on the body. This alarm can be transmitted in the form of a radio transmission to a rescue center.

However, the above systems can lead to acceptance problems on the part of the persons concerned, since the permanent wearing of the devices for these persons is cumbersome and inconvenient. In addition, especially with older people has the problem that they may simply forgot to re-create the appropriate equipment, for example, after a bath. Fall detection is also possible via sensors introduced into a floor or mats equipped with sensors, however, an implementation of such a system is very complicated and expensive.

Another class of methods for fall detection and detection of special emergencies are optically based methods. Here, a permanent optical monitoring of the person is performed as well as an optical fall detection / emergency detection, the classification using common classification algorithms such as Bayes classifiers, neural networks or principal component analysis of a recorded Event and triggers an alarm upon detection of a fall / distress and sends it to a higher-level instance. US20060145874A1 relates to an optically based method, wherein the image processing algorithms in particular determine the percentage of a person lying on the ground, the position of the person, as well as the size of the person. In addition, the speed and acceleration of a falling person is determined. US20030058341A1 relates to an optically based method in which an object is released from a scene, this is detected and analyzed. In doing so, certain properties of the object are analyzed on the basis of comparison values with specific events. In particular, an event classification using geometric primitives is possible, as described in Chia-Wen Lin, Zhi-Hong Ling, Yeng-Cheng Chang and Chung J. Kuo "Compressed-Domain Case Incident Detection for Intelligent Homecare", submitted to Journal of VLSI Signal Processing. US 20050146605 A1 relates to an optical monitoring method with geometric primitives and subsequent event analysis using event discriminators, whereby an alarm can be triggered depending on the detected event. Procedures that involve permanent video surveillance of a person, however, interfere with the personal rights of the person and can therefore lead to acceptance problems. Offenbarune of the invention

It is therefore an object of the invention to provide a method and system for fall detection that is reliable and works safely. This object is achieved by a method and system according to independent claims 1 and 10. The dependent claims 2 to 9 and 11 to 12 relate to particular embodiments.

Here is a space by means of a passive optical sensor, such as a video or. Infrared camera monitors. Furthermore, an evaluation of the recorded images with respect to the detection of a fall or a special emergency situation, but the image data are initially transmitted only alienated to an authorized receiver triggering an alarm upon detection of a fall. In this way, the right to one's own image of the monitored person is preserved, since not permanently images of this person, but only alienated images are transmitted upon detection of a fall.

If a fall or a special emergency situation is detected, a return channel is first activated, which serves to verify the detected situation. The supervised person is enabled to classify a triggered alarm as a false alarm.

Only when after a certain time an alarm is not classified as a false alarm by the monitored person, a decryption of the transmitted optical signal, so that the monitored person is displayed. This pictorial representation of the situation and person would reveal at the moment also an arriving rescue service.

The alienation of the image data is preferably carried out by means of geometric primitives. For example, those present in a captured image Each replaced by one or more geometric primitives. Preferably, each person present in the captured image is treated and displayed as at least one geometric primitive. Preferably, the fall detection is determined on the basis of the geometric primitive or for each person.

If a fall or a helplessness of a fallen person is detected in the unfamiliar image, the authorized recipient may initiate further rescue measures, such as request rescue services or instruct medical personnel to take care of the fallen person.

The essence of the invention lies in a method and system for detecting a particular emergency especially a fall detection, wherein initially alienated and possibly encrypted image data of this person and / or the room to an authorized during the permanent optical monitoring of a person only in emergency or fall detection Instance are triggered by triggering an alarm. Only after a return channel has been formed and if the monitored person fails to classify the alarm as a false alarm will an unaltered image of the situation or of the person be transmitted to an authorized entity. This method and system has the advantage that it does not restrict the physical freedom of action of a person to be monitored, for which the person is comfortable and the person's right to his own image as an outflow of the general personality rights and also inexpensive and easy to handle for medical personnel and / or relatives is.

Brief description of the figures

The method and the associated system of the invention will be described in more detail below with reference to exemplary embodiments and the drawings.

Show it:

1 shows a system for detecting a special emergency or fall detection with a server and a universal standard gateway and an array of optical sensors, Figure 2 shows a system for detecting a particular emergency or fall detection with a health box and a health platform and an array of optical sensors, and Figure 3 shows a system for detecting a special emergency or fall detection with a LogicHealth box and a health platform wherein the algorithms for video analysis and fall detection are implemented in an optical sensor.

Detailed description of the embodiments

Figure 1 shows an embodiment of the system. In this case, images of a person 118 to be monitored are permanently recorded by means of a number of optical sensors 11 which have individual camera sensors 11.1 to 11.1. The camera sensors 11.1 to 11.n can be arranged in a space to be monitored such that the person to be monitored is detected by at least one camera sensor at all times. The optical sensors 11 are connected to a server 13 via an interface 19. In the server 13 there is a logic 12 in which the algorithms for decomposing the images recorded by means of the sensors 11 into geometric primitives as well as algorithms for fall detection or for detecting a special emergency situation are implemented. If a fall or an emergency situation is detected in this case, then an alienated image, which by means of a device and method for encryption and authentication, which may include a firewall 15, via a TCP / IP interface 17 and the Internet 113 to an authorized circle of people 117 if necessary ., transmitted via a universal standard gateway 114, wherein the transmission can be wireless 115 or wired 116 bidirectional.

If a particular emergency or a fall is detected, an alarm is triggered which the monitored person 118 can invalidate (classify as a false alarm). For this purpose, a return channel is established via a data or telephone connection between the monitored room and the authorized receiver. If, after a certain period of time, an alarm is not classified as a false alarm, then an unaltered transmission of the picture or a sequence of pictures to the authorized group of people takes place. The wired 116 or wireless 115 transmission is also encrypted. Another possibility for forming a return channel is additional given by an acoustic device 14 and a sound card integrated into the system 18. Here, an acoustic connection between the authorized circle of people 117 via the Internet 113 and the server 13 via the sound card 18 and the acoustic device 14, such as a speaker and a microphone made, which may be bidirectional. To the server 13, which is executed in this embodiment as a PC with additional logic, more sensors 112 or user menus 111 may be connected via the interface 19.

The logic 12 in the embodiment of Figure 1 includes the software for performing video image analysis, crash detection analysis and encryption, and data communication.

FIG. 2 shows a further embodiment. In this embodiment, the server 13 of Figure 1 is implemented as a so-called Healthbox 26, i. a device for monitoring the health of a person. The image processing and the data processing of all health sensors and the data encryption takes place in a Logic HealthBox 25, which is embedded in the Healthbox 26. The authentication and routing of the information takes place in an external HealthPlatform 27, which can be a system for monitoring a large number of persons. HealthBox 26 and HealthPlatform 27 are interconnected via the Internet 113. The forwarding of the alarm or the alienated and unrecognized in a further stage images is encrypted either wirelessly 115, for example. On smartphones or wired 116, for example, to a telephone or external computer to an authorized circle 117 people. The optical sensors 11.1 to 1 ln are over an interface 19 connected to the Healthbox 26. The camera sensors only record the pictures. The server of the first embodiment of FIG. 1 with the further devices is implemented in the form of the healthbox 26 in this second embodiment. The health platform 27 of this second embodiment also has the functionality of the Universal Standard Gateway 114 of the first embodiment. As for the other elements of this embodiment, reference is made to FIG.

The logic HealthBox 25 in the embodiment of Figure 2 includes the software for performing the video image analysis, the fall detection analysis, the Data processing of all sensors, especially health sensors and encryption as well as data communication.

Figure 3 shows a third embodiment of the system. The optical sensors are in the form of a master 30 and a plurality of slaves 41, 42. The optical sensor operating as master 30 has a camera part 31, an internal interface 32, a smart camera logic 33 and an interface 34. In this embodiment, in contrast to the first and second embodiments, the software and the associated evaluation unit, which implements the algorithms for video analysis and emergency detection, are integrated in the SmartKamera logic 33 of the master optical sensor. The smart camera logic 33 is internally connected to the camera part 31 via the interface 32. The SmartKamera logic 33 is further connected via the interface 34 with the other operating as a slave optical sensors 41, 42 of the system and the interface 19 with the Healthbox 26. With regard to the remaining elements and functionality of this embodiment, reference is made to FIG. 1 and FIG.

The smart camera logic 33 in the embodiment of Figure 3 includes the software for performing the video image analysis and the fall detection analysis. The logic HealthBox 35 has the remaining functions contained in the logic HealthBox 25 of the embodiment of FIG. This includes the software for the data processing of all sensors, in particular health sensors and the encryption as well as the data communication. The invention will be explained in more detail below with reference to functions and properties.

The optical sensors are preferably to be placed in the room in such a way that optimum coverage of the space to be monitored, even in winding rooms, is achieved. Optionally, a space area is to be covered by more than one optical sensor. Preferably, an active energy management is implemented in this case, so that only optical sensors are in active mode that are currently required. The optical sensors are designed such that they are robust to variable light conditions such as reflections and backlight. In addition, these can be designed in the form of infrared sensors to detect falls and emergencies even in the dark can.

The fall detection algorithms are implemented in such a way that they can detect any types of crashes, such as collapse, heavy fall, and sustained fall. These types of falls are well defined in advance and are part of a library of activated fall detection algorithms ("patterns.") Emergencies are identified primarily by algorithms that evaluate the static state of a person relative to the home environment and, in particular, detect an abnormal static condition ,

The recognition algorithms are implemented in such a way that they are robust with respect to other persons or animals in the room as well as with respect to the positions of the optical sensors and obscuring the person to be monitored by furniture and / or other persons or animals. Likewise, they are able to classify rapid movements of objects not as collapses. The recognition algorithms include methods of extracting a person from the background and vectorizing and decomposing the image of a person into geometric primitives and determining the location of them in space either in the server or in the optical sensor. Fall detection uses the dynamics of a fall. It analyzes how fast the position of these geometric primitives changes within a sequence of images. The image processing can be done centrally in a dedicated Healthbox and be connected to a computer cloud to minimize maintenance and to easily expand the evaluation patterns.

The system is designed so that within a certain period all persons of an authorized group of persons are informed. This authorized group of persons only has access to personal data and status information about the person to be monitored. The system is designed in such a way that the authorized group of persons can be supplemented by additional persons or reduced by certain persons. In case of a validated alarm, the system will provide the authorized circle of people with the time of the fall, the place of the fall, a still image and possibly a video. To protect privacy, the system displays at various places when video or video material leaves the room to be monitored. The recording of image or video data of the person to be monitored preferably takes place only for a period which is absolutely necessary for the functionality of the fall detection system. The security level of the telecommunication is preferably scalable, so that between different security levels of the system from weakly protected (low end security level) to strongly protected (high end security level) can be selected.

The system may additionally contain suitable administration, billing, routing and customer administration functionalities on the platform. In particular, the system is preferably designed so that it can be easily maintained from the outside by a device for the automatic upgrade of firmware and drivers as well as the algorithms.

The system is preferably designed to be scalable so that additional optical and non-optical sensors can be added or removed to the system to complement the functional properties and performance of the system.

The overall system is designed in such a way that it visually and by design avoids any impression of surveillance and inconspicuously blends into the living ambience. The invention has been explained in more detail with reference to examples and the figures, this representation is not intended to limit the invention. It is understood that those skilled in the art can make changes and modifications without departing from the scope of the following claims. In particular, the invention includes embodiments with any combination of features of the various embodiments described herein.

Claims

claims

1. A method for detecting emergency situations and fall detection and passing an alarm with the following steps:

a) monitoring a room by means of at least one optical sensor, such as a video or infrared camera,

b) detecting an emergency situation or a fall of a person in the monitored space by means of evaluating the images recorded with the optical sensor, c) alienating the image data, and

d) triggering an alarm and sending the alienated image data when an emergency / fall of a person in the monitored room is detected, to at least one authorized receiver.

2. The method of claim 1, further comprising the steps of:

Activating a return channel via which a data / telephone connection between the monitored room and the receiver is set up, wherein the triggered alarm can be identified as a false alarm via the return channel.

3. The method according to claim 1 or 2, with the further steps:

Encrypted transmission of at least one still image and / or video sequence recorded by the optical sensor in the space to be monitored to the at least one receiver, after lapse of a predetermined time after the alarm is triggered, or in the event that the alarm not marked as false alarm.

4. The method of claim 1, wherein the image data is alienated by transforming at least the persons recorded in the monitored space into geometric primitives and transmitting these geometric primitives to the at least one receiver in the first stage of the alarm.

5. The method according to any one of claims 1 to 4 wherein simultaneously with the alienated image data, further data in encrypted form to at least one Receivers are transmitted, which further data comprise at least one still image and / or video sequence of a monitored space and wherein preferably the encrypted data can be decrypted and displayed by the receiver only after confirming the alarm.

6. The method according to any one of claims 1 to 5, with the further steps:

Storing the images and / or a video sequence of a certain period of time before and after detecting a fall in a memory of the optical sensor and transmitting the stored images and / or the video sequence as alienated image data and / or as encrypted data to at least one of the receivers.

7. The method according to any one of claims 1 to 6, with the further steps:

Submit the alert and any additional information provided to at least one authorized recipient selected on a case-by-case basis from authorized and authorized individuals such as individuals, nurses, and ambulance through a dedicated management and delivery platform.

8. The method according to any one of claims 1 to 7 with the further steps:

Detecting an emergency situation or a fall of a person in the monitored space by means of non-optical sensors, preferably acceleration sensors, which carries the person on the body and / or sensors on or in the floor of the monitored room and / or an alarm call device that can be activated by a person and / or an acoustic device. ,

9. The method according to any one of claims 1-8, wherein a static emergency situation using a house emergency call or the evaluation of acoustic signals is detected.

1 O.Verfahren according to one of claims 1 to 9, with the further steps:

Indicating whether alienated image data and / or encrypted data has been sent and / or storing events, and preferably protecting the entire facility against abusive data and information access.

11. A system for detecting a particular emergency or fall detection and passing an alarm, comprising:

a. at least one optical sensor, such as a video or infrared camera, for monitoring a room,

b. Detection means for detecting a particular emergency or a fall of a person in the monitored space and evaluation means for evaluating the images recorded with the optical sensor, c. Alienation means for alienating the image data and

d. Means for triggering an alarm and sending the alienated image data when a fall of a person in the monitored room is detected to at least one authorized receiver. 12.The system of claim 10, comprising:

Activation means for activating a return channel over which a data / telephone connection between the monitored room and the receiver is established, wherein over the return channel optionally the triggered alarm can be marked as false alarm.

The system of claim 10 or 11, comprising:

Encryption means for encrypted transmission of at least one still image and / or a video sequence.