WO2012002904A1

WO2012002904A1 - Device and method for detection of abnormal spatial states of a human body

Info

Publication number: WO2012002904A1
Application number: PCT/SE2011/050901
Authority: WO
Inventors: Emil HÄLLSTIG
Original assignee: Fotonic I Norden Ab
Priority date: 2010-07-02
Filing date: 2011-07-04
Publication date: 2012-01-05

Abstract

The present invention relates to a method and device (100) for privacy- preserving non-invasive detection of an abnormal spatial state of a human body within a predefined space. The device (100) comprises an illumination unit (110), a sensor (130), a processing unit (140) and a memory unit ( 150), The illumination unit (1 10) is configured to emit light into the predefined space within a predetermined wavelength range and modulation scheme and the sensor (130) is configured to receive a reflected signal from the predefined space as a response to the emitted light and to convert variations in the reflected signal into a snapshot (170) representing a distance image of the predefined space. The processing unit (140) is adapted and configured to detect if a human body (50) is present in the predefined space and if detected identify and monitor a least one reference point (805) of the human body and trigger an alarm device (30) if the pattern of the at least one monitored reference point (805) is indicative of the human body (50) experiencing an abnormal spatial state.

Description

DEVICE AND METHOD FOR DETECTION OF ABNORMAL

SPATIAL STATES OF A HUMAN BODY

Techekal Field

The present invention relates to a device and method for detection of abnormal spatial states of a human body, especially non-invasive detection in a privacy-preserving way.

Background

The increasing average life expectancy and the increasing number of senior citizens constitute a challenge for our society in terms of increasing needs for

monitoring, assistance, care and appropriate housing options, in later life people experience changes and problems, often interconnected, that strongly affect their situation, such as retirement, relocation and health problems. Technology can contribute with solutions to these new problems; but there are some important ethical issues to consider, such as autonomy, integrity and quality of life. The increasing and aging population with increasing needs for security and care are just!y calling for integrity and a dignified life. Moreover, they are calling for a dignified death. Many elderly people are dreading the risk of being left undetected for days or weeks. Meeting these expectations at a reasonable cost, without introducing other, perhaps unforeseeable disadvantages, is a huge and complex problem.

Many people have, by age or sickness, an increased risk of fall injuries. One third of all people over 65 living at home will have a fall accident each year. In nursing homes and hospitals the fall accident frequency is ten-folded. According to e.g. British and Swedish studies some 10 % of these fall accidents cause a serious injur}'', and 1 -2 % causes hip fractures and/or concussion. As an example, 1260 fall accidents can be expected to occur annually at a hospital with 800 beds. Besides the human suffering caused by these accidents, they also entail heavy costs for treatment. For a person living at home, a fall that is not in it self fatal may all the same ultimately cause great suffering and death, if the fall has prevented the person from calling for help. To mitigate this risk several solutions for fall detection have been presented. Many of them, such a for instance the solution described in US2G09073991 A, are based on accelerometers that the care-taker is meant, to wear. Other solutions advocate subcutaneous accelerometer implants. Still others are based on weight sensors mounted on rugs. The wearable accelerometer is easily forgotten or lost by a care-taker, which may have impaired vision or suffer from memory loss or dementia. Invasive surgery is both improper, and risky. Rugs in themselves constitute a risk, in that caretakers may- trip or slip on them, and therefore, a weight-sensitive rug is not a good solution, In conclusion, these solutions are invasive, impractical, unsafe, unreliable, and

inappropriate and have an unacceptably high error rate.

Some solutions rely on camera surveillance and advanced image processing. The fall detection solution of US 20090121881 is divided into two steps; finding the person on the fl oor and examining the way in which the person ended up on the floor, including location, direction, velocity and acceleration. When the first step indicates that the person is on the floor, data for a time period before and after the indication may he analyzed. One major drawback of this solution is that the accuracy is affected by aspects like the inclination of the camera, and the relative direction of the fall. Moreover, having an advanced video camera in a bedroom or a bathroom, that constantly records and analyses a caretaker's every activity is a serious violation of the personal integrity, and will be unacceptable to the absolute majority of caretakers.

US7567200B features analysis of a. reflected radar signal that determines a care-taker's moving body segment and its distance to a floor. The contraption requires several receivers, a transmitter with a certain baseline relative the receivers, and an advanced signal processor. Furlhennore, in order for the signal processing to work, the breathing motion must be detectable from a receiver. Consequently, the detector as described in US7567200B is not capable of detecting a person with respiratory or cardiac arrest, and may likely have difficulty detecting a person having the back turned against the receivers.

Finally US 20020044682 describes a method and an apparatus for analyzing an orientation of a subject, such as a human body. An image of the volume in which the subject might be present is captured by a camera capable of taking stereoscopic images, by taking a left image and a right image, The left and right images are then sent to a appropriate image processing device, such as a computer, in order to create a three- dimensional image of the subject. This three-dimensional image may be further analyzed in order to determine an undesirable orientation of the subject being monitored. The necessity of using two images to create a three-dimensional image and the external computing resources renders this solution complex and not as fast as desirable.

Thus there is a need for a method and a device that in a simple and reliable way are able to detect an abnormal spatial state of a human body in real- time, without the need of complex signal processing, or space and time requiring triangulation or scanning equipment.

Summary

The object of the present invention is to obviate at least some of the above disadvantages and provide improved methods, apparatuses and computer media products avoiding the above mentioned drawbacks.

According to a first aspect of the present invention this is accomplished by a device for detection of an abnormal spatial state of a human body within a predefined space. The device comprises an illumination unit, a sensor, a processing unit and a memory unit. The illumination unit is configured to emit light into the predefined space within a predetermined wavelength range and modulation scheme and the sensor is configured to receive a reflected signal from the predefined space as a response to the emitted light and to convert variations in the reflected signal into a snapshot

representing a distance image of the predefined space. The processing unit is adapted and configured to detect if a human body is present in the predefined space and if detected identify and monitor a least one reference point of the human body and trigger an alarm device if the pattern of the at least one monitored reference point is indicative of the human body experiencing an abnormal spatial state.

In a preferred embodiment of the present invention present invention the processing unit is further adapted and configured to detect presence of the human body when a model of a human body can be fitted to the snapshot. In yet another preferred embodiment of the present invention the processing unit is adapted and configured to identify and monitor a set of points, comprising the at least one reference point and trigger the alarm device if the pattern of the set of points s indicative of the human body experiencing an abnormal spatial state,

In another embodiment the processing unit is adapted and configured to indi cate an abnormal state of the human body if the at least one reference point or the set of points is/are in a certain position relative a predefined level and/or if the at least one reference point or the set of points is/are moving above a predefined velocity.

According to a second aspect of the present invention there is accomplished a method for detection of an abnormal spatial state of a human body within a predefined space. The method is performed with a device comprising an illumination unit, a sensor, a processing unit and a memory unit. The method comprises following steps:

emitting light from the illumination unit into the predefined space within a predetermined wavelength range and modulation scheme,

receiving a signal reflected from the predefined space as a response to the emitted light,

converting variations in the reflected signal into a snapshot representing a distance image of the predefined space,

detecting if a human body is present in the predefined space,

identifying and monitoring a least one reference point of the human body and triggering an alarm device if the pa ttern of the at least one monitored reference point is indicative of the human body experiencing an abnormal spatial state.

In a preferred embodiment of the present invention the step of detecting presence of the human body comprises fitting a model of a human body into the snapshot.

In yet another preferred embodiment of the present invention the method further comprises the steps of identifying and monitoring a set of points, comprising the at least one reference point and triggering the alarm device if the pattern of the set of points is indicative of the human body experiencing an abnormal spatial state. In another preferred embodiment the method comprises the step of indicating an abnormal state of the human body if the at least one reference point or the set of points is/are in a certain position relative a predefined level and/or if the at least one reference point or the set of points is/are moving above a predefined velocity.

In a preferred embodiment of the present invention the reference point is a point of the head of the human body,

According to a third aspect of the present invention a computer program is accomplished comprising code means for performing the method according to the second aspect,

According to a fourth aspect of the present invention a computer program product is accomplished comprising program code means stored on a computer readable medium for performing the method according to the second aspect, when said product is run on a computer.

With the device and the method above the detection of a human body and if it is experiencing an abnormal spatial state is instant, secure, accurate and reliable. The device is compact, convenient, and non-invasive in that it does not. draw attention to itself and is also privacy-preserving. The latter comes from the fact that while the posture and location of a persons body can be instantly and exactly determined the present invention, personal characteristics that enable identification, such as hair color, facial features, facial expressions and voice may not be instantly detected by the time- of-flight principle in itself. Further, it does not draw attention to a monitored person.

Brief Description of the Drawings

In order to explain the invention in more detail an embodiment of the present invention will be described in detail below, reference being made to the accompanying drawings, in which

Figure 1 a is a schematic view of a scene being monitored,

Figure lb is a schematic model of a device according to the present invention, Figure 2 is a visualization of a digitized representation of a monitored body and Figure 3 is a flow chart outlining a method according to the present invention. Detailed Description

Since the present invention as a part of its solution uses a tirae-of flight sensor, hereinafter referred as a TOF sensor, the principles of the TOP sensor will first be described.

A TOF sensor creates distance data with the help of light signals, usually pulses that are switched on for a very short time and are illuminating a predefined space, i.e. a detection scene. The signal is then scattered, i.e. reflected, by surfaces and objects in the scene. The light is collected onto a sensor plane of the TOF sensor. Light incident on the sensor plane from farther points of the object will be delayed compared to light reflected from nearer points. The delay has the magnitude of nanoseconds only. As each pixel of the sensor plane represents one unique point of the object, the spatial delay distribution represents a three-dimensional measurement of the object.

The pulse power and frequency of the illumination determines the maximum range the sensor can handle, and the range uncertainty. The illumination unit is therefore an important part of the TOF sensor. The light has to be modulated with a very high speed, up to 100 MHz, and only special LED's (Light Emitting Diodes) or lasers can generate such short pulses. A TOF sensor may cover ranges of a few meters up to over 60 m. The distance resolution of a TOF sensor may be less than 1 cm, the time resolution may be more than 100 images per second. The illumination unit may use infrared or near-infrared (NIR) light to make the illumination privacy preserving and non-invasive.

An optical unit of the TOF sensor comprises a lens and may also comprise a band pass filter. The lens gathers the reflected light and images the scene onto the sensor plane. The optical band pass filter only allows the light with the same

wavelength as the illumination unit to pass through, and thus suppresses all background light.

The sensor plane comprises an array of pixels. Each pixel may comprise two areas with alternating activation, a so called two-tap phase lock detection principle. The differences between intensities detected in the two areas correlates to the distance from the sensor pixel to one particular point of the detected scene. There is no need to use mechanical scanning, with prisms or mirrors, for a TOF sensor, which makes the construction robust. Furthermore, no electronics for mixing the received and reference signals are needed outside the TOF sensor itself. The distance is calculated directly in the TOF sensor and may then provide a distance image over a USB or Ethernet interface, in context of this application distance image is an image comprising range information, i.e. information about the distances from the sensor plane of the TOF sensor to different object in the detection scene.

In contrast to stereo vision or triangulation systems, the TOF sensor has an advantage in that, it is very compact, the illumination unit may be placed just next to the lens, whereas other systems need a certain minimum base line. In contrast to laser scanning systems, no moving mechanical parts are needed. Furthermore, the distance algorithm in a TOF sensor is very efficient, and it is easy to extract the distance information from the output signals of the TOF sensor. Therefore this task uses only a small amount of processing power, again in contrast to stereo vision cameras or the above mentioned radar application, where complex correlation algorithms have to be implemented.

After the distance data has been extracted, object detection, for example, is also easy to carry out because the algorithms are relatively insensitive to the reflection from the object, whereas previously known detectors and detection methods have problems with uniform objects or repeatable patterns on the object. Furthermore, TOF sensors are able to measure the distances within a complete scene with one shot. TOF sensors reach up to 100 frames per second, and are therefore suited to be used in real-time

applications

An embodiment of a detection device according to the present invention will now be described in relation to Figure la and lb. The detection device 100 is used for detection of an abnormal spatial state of a person 50, i.e. a body in a position or in motion which does not occur under normal circumstances or that is not desirable. The detection device 100 may be mounted in the ceiling or some other suitable place, as long as there is a line of sight between the detection device 100 and the detection scene to be monitored. The detection device 100 comprises an illumination unit 110, an optical unit 120, a sensor 130 having a sensor plane 135, a processing unit 140, a memory unit 150 and an interface 160. A driver electronics unit 190 provides the illumination unit 110 and the sensor 130 with power.

The illumination unit 1 10 is adapted and configured to send out a detection signal 10, i.e. emit light into the detection scene or a predefined space with a predetermined wavelength range and modulation scheme. The modulation scheme usually comprises pulses and the wavelength range may correspond to 1R or NIR light in order to obtain inconspicuous monitoring. By using 1R or NIR light monitoring will be unnoticeable even during the night and will therefore not disturb the caretaker 50. Using non-visible light wiil further prevent outside spectators from being able to se the caretaker, which of course is an advantage.

The sensor plane 135 comprises an array of pixels. Each pixel in the sensor plane 135 is adapted to receive a signal 20 reflected from one point of an object in the illuminated detection scene. The sensor 130 is configured to simultaneously convert variations in the reflected signal 20 into a snapshot 170 representing a distance image, i.e. as mentioned above an image comprising range information, using properties of the detection signal 10 as reference. A set 1 80 of one or many snapshots 170 may form the basis for a digitized representation of the entire scene as a relief map.

The processing unit 140 is adapted and configured to detect if a human body 50 is present in the predefined space and trigger an alarm device 30 if the there is an indication that the human body 50 is experiencing an abnormal spatial state, which will be described closer below. The detection of the human body 50 may be done in a number of ways. When the processing unit 140 screens the distance image it may determine whether a model of a human body can be fitted into that image, and if so confirm human presence. The model preferably is a three-dimensional model, but it may also be a two-dimensional model. The fitting can be made according to some known available algorithm.

The device 100 also comprises a memory 150 configured and adapted for intermediate storing of data prior to, during, or after analysis of data received from the sensor plane 135 or from the processing unit 140. Generic information about postures, defined as combinations of reference points, and patterns classified as abnormal for human beings can be pre-stored in the memory ] 50. Information can also be generated during real-time situations and stored in the memory 150, so that the device 100 may be individually trained.

When human presence is established, the device 100 monitors the human body 50 found during screening and the processing unit 140 identifies and monitors at least one reference point 805 (see figure 2) on the human body in order to find a pattern indicative of the human body 50 experiencing an abnormal spatial state. If an abnormal spatial state is detected the processing unit 140 triggers an alarm device 30 so that an alert 35 goes off. The alert 35 may be generated within the device 100, but is typically intended for staff 37 located in separate premises; therefore it may be generated in a remote alarm device 30 e.g. as an acoustic signal 35 from a horn, a message on a pager, a special phone ringing signal to members of staff, a Hashing light on a control panel, or any combination.

The interface unit 160 is adapted and configured to allow the device 100 to interact and communicate with a remote alarm device 30, and to send a trigger signal 165. It is further possible to let the remote alarm device 30, or the trigger signal 165 from the processing unit 140 itself, trigger one or many external devices 40 providing additional functions 42. For instance, as the special ring signal 35 is triggered, a phone call 42 Is set up between the staff phone and a speaker phone 40 in the monitored room, allowing staff 37 to speak to the monitored person 50. Advantages of embodiments like these are that staff 37 can get more information on what has actually happened, so as to prepare correctly, the monitored person 50 gets a confirmation that help is on Its way, and may be comforted by this knowledge, or by talking to staff 37. In the event that the light is turned on in the monitored room, the alert 35 may be combined with turning a lighting device 40 on, so that a fallen person does not have to lie in the dark.

Two snapshots may be compared and the deviation between the two will represent the projected volume of an object added to the scene at a point in time in between creation of the first and second snapshot. It is also possible to analyze special features like shape and size in order to recognize a human body on the basis of one snapshot 170 only, it is further possible to use certain algorithms onto a digitized projection of a person to create a true digitized representation of that person, either as a solid body, or as a three-dimensional body surface. This digitized representation is further related to some coordinate system. The native coordinate system Q (ξ, υ, R) is illustrated in figure 1. Each point of the representation corresponds to an individual pixel, ς and denotes where the pixel is located in the grid, and R denotes the distance from the pixel to the point. This system however is not very intuitive, in order to optimize the solution for in-door applications the data may therefore be geometrically transformed into a Cartesian system. That way, the x- and y-coordinates can describe the lateral dimensions and z describes the vertical dimension, the height of each point above a certain base level, which may be equal to the floor, or some other threshold plane.

Figure 2 shows a digitized representation 800 of a human body. As mentioned above the detector 100 is adapted and configured to identify and monitor a certain reference point 805, or set of multiple reference points. Reference points of particular interest according to the present invention are any points 810 in the far end of a limb, such as a head, a leg or an arm, the topmost poi nt 820 of the human body, regardless of limb, some point 830 of the head, the area or volume centroid 840 of a body part such as the head or the trunk, or of the entire body, or another easily delected body part 850, such as nose, ear, mouth, elbow, knee, shoulder, or other joint. As an alternative to points, distances, vectors, surfaces or volumes may be used, bearing in mind that a digitized surface or volume is in effect a set of points. Yet an alternative is to constantly comparing successive snapshots to create velocity vectors.

The detector 100 is adapted to identify events related to spatial or temporal characteristics of the monitored reference points, so called patterns. Events of particular interest to embodiments of the present invention are: a reference point 805 being in a certain position relative a predefined level, particularly the vertical dimension, or relative some other reference point; a reference point moving above a predefined velocity, or with a certain acceleration; movement of the entire body; movement of a body part; relative body part movement or motion pattern. One event may be relative movement of body parts indicative of a state of elevated emergency, such as an epileptic seizure etc. Yet an alternative is to calculate projected volume downward, and define a minimum volume as the pattern.

In order to detect an abnormal spatial state of a human body, the detector 100 may be adapted and configured to monitor the pattern of the head. A head moving at high speed is almost always an unintentional action, which indicates that the monitored person 50 has lost control over his or her body, i.e. experiences and abnormal spatial state. Based on this event or pattern, a fall can be predicted to occur, if a reference point 830 on the head is positioned within a certain range from the floor it indicates a fall. If the point 830 is situated at a lower level than e.g. some point on the shoulder 850 or trunk 840, this in itself may constitute an abnormal spatial state. An abnormal spatial state can thus be predicted, indicated or established by detecting one or many reference points 805 showing a pattern, or combination of patterns. Certain postures may be allowed for a short period, i.e. will not trigger an alarm immediately, but are considered abnormal if upheld for a longer period. Certain postures may be allowed in certain areas. For instance, a pattern corresponding to a person lying flat may be disallowed in general, except for a certain area in the x-y dimensions, where the bed is situated.

For the sake of clarity, it should be emphasized that all points 810-850 may be reference points, and that the set of points 860 may comprise any combination of reference points.

To better understand the present invention a method for detection of abnormal spatial state of a human body, such as e.g. a fall, or a broken limb will be described. The method is practiced in a device 100 as described above. The method is shown in figure 3. The method comprises the steps of emitting light 200 from the illumination unit 110 into the predefined space. The light is emitted within a predetermined wavelength range and modulation scheme. The method further comprises receiving 300 a signal reflected from the predefined space as a response to the emitted light and converting 400 variations in the reflected signal into a snapshot 170 representing a distance image of the predefined space. The snapshot 170 is then used to detect 500 if a human body 50 is present in the detection scene or not. If no human body 50 is detected the detection device 100 waits until a human body is detected.

When a human body has been detected in step 500 a reference point 805 of the human body will be identified and monitored in step 600. If the monitored reference point 805 shows or has a pattern that is indicative of the human body 50 experiencing an abnormal spatial state an alarm device 30 will be triggered in step 700.

In the receiving step 200, the processing unit 140 is collecting a stream or set 180 of real-time snapshots comprising range information. The rate of snapshots does not have to be high. For many purposes a rate of e.g. one snapshot per minute may be sufficient. Such a relatively low rate will save energy. The frames may be analyzed in the processing unit 140 directly upon arrival from the sensor plane 135, or stored intermediately, or buffered, in the memory unit 150.

In the detection step 500, the processing unit 140 is screening the stream or set 180 of snapshots for human presence. This step may comprise the further steps of determining whether a model of a human body can be fitted to data emanating from a distance image, and if so establishing human presence. The model may show various levels of detail, depending on the specification and purpose.

If a pattern, indicating an abnormal spatial state is recognized in the monitoring step 600 an alert triggering signal 165 is generated in the triggering step 700 as mentioned above. Via the interface 160 the alert triggering signal 165 is sent to an external device 30, which upon reception of the signal 165 sets off the alert 35.

The above mentioned criteria for fall detection may he used and combined according to the conditions under which the monitoring is taking place. In most situations, the fall detector will be monitoring a single person only, or just a few people. The detector can therefore advantageously be trained to recognize a particular human body or a set of human bodies. It is important to remember, though, that no

identification of a human being is needed, only an anonymous body scanning. Based on this scanning a set of parameters are calculated, which is then are used by the fall detector when monitoring the human body. These parameter sets may comprise height, shape of head, length of arms, length of legs, width across the shoulders, chest width, hip measurements and/or body volume. The parameters may be absolute or relative some other parameter.

The aspects of the present invention give many advantages. Firstly, the monitoring is invisible and inconspicuous and will not bother a caretaker. The system can detect all sorts of abnormal states, from absence of movement of chest, indicating respiratory problems, to a body positioned on the floor, whether it is moving or not. It can detect not only falls and people being positioned on the floor but also epileptic seizures, spasms or convulsive fits. It may also detect people with respiratory or cardiac arrest, including a deceased person. It is ossible: to train the sysfeoi to react ^'to. eertain tions^' including ^' artic lar m emen patterns iw ieh may M-ttoroai for west eci le_? but !br ¾ certain. i.n hid.«aj a sign thai sometliing is wrong; to intentional signaling ii05« a caretaker in a. situation iMt pre'veriis calling for assisiance ia eonventioual ways, Tlie system i very accurate la terms Pfpc ftdn. and motion_* yet discriminates against facial f½ture¾, racial expressions and other parameters that en ble -identification, or reading -State of mind, mid 'thus integrity preserving,

Claims

1. A device (100) for detection of an abnormal spatial state of a human body within a predefined space, comprising an illumination unit (1 10), a sensor (130), a processing unit (140) and a memory unit ( 150), wherein the illumination unit (1 10) is configured to emit light into the predefined space within a predetermined wavelength range and modulation scheme, the sensor (130) is configured to receive a reflected signal from the predefined space as a response to the emitted light and to convert variations in the reflected signal into a snapshot (170) representing a distance image of the predefined space and the processing unit (140) is adapted and configured to detect if a human body (50) is present in the predefined space, identify and monitor a least one reference point (805) of the human body and trigger an alarm device (30) if the pattern of the at least one monitored reference point (805) is indicative of the human body (50) experiencing an abnormal spatial state.

2. The device (100) according to claim 1, wherein the processing unit (140) further is adapted and configured to detect presence of the human body (50) when a model of a human body can be fitted to the snapshot.

3. The device (100) according to claim 1 or 2, wherein, the processing unit (140) further is adapted and configured to identify and monitor a set of points (860), comprising the at least one reference point (805) and trigger the alarm device (30) if the pattern of the set of points (860) is indicative of the human body (50) experiencing an abnormal spatial state,

4. The device (100) according to claim 3, wherein the processing unit (140) further is adapted and configured to indicate an abnormal state of the human body (50) if the at least one reference point (805) or the set of points (860) is/are in a certain position relative a predefined level.

5. The device (100) according to claim 2 or 3, wherein the processing unit (140) further is adapted and configured to indicate an abnormal state of the human body (50) if the at least one reference point (805) or the set of points (860) is/are moving above a predefined velocity.

6. A method for detection of an abnormal spatial state of a human body within a predefined space, said method being performed by a device (100) comprising an illumination unit (110), a sensor (130), a processing unit (140) and a memory unit (150) and the following steps:

emitting light from the illumination unit (110) into the predefined space within a predetermined wavelength range and modulation scheme,

converting variations in the reflected signal into a snapshot (170) representing a distance image of the predefined space,

detecting if a human body (50) is present in the predefined space,

identifying and monitoring a least one reference point (805) of the human body and

triggering an alarm device (30) if the pattern of the at least one monitored reference point (805) is indicati ve of the human body (50) experiencing an abnormal spatial state.

7. The method according to claim 6, wherein the step of detecting the presence of the human body (50) comprises fitting a model of a human body into the snapshot (170).

8. The method according to claim 6 or 7, further comprising the steps of identifying and monitoring a set of points (860), comprising the at least one reference point (805) and triggering the alarm device (30) if the pattern of the set of points (860) is indicative of the human body (50) experiencing an abnormal spatial state.

9. The method according to claim 8, further comprising the step of indicating a abiiormal state of the human body (50) if the at least one reference point (805) or the set of points (860) is/are in a certain position rel ative a predefined level,

10. The method according to claim 8 or 9, further comprising the step of indicating an abnormal state of the human body (50) if the at least one reference point (805) or the set of points (860) is/are moving above a predefined velocity.

11. The method according any of claims 6 to 10, wherein a point (830) of the head is the reference point.

1.2, A. computer

method of any of claims 6 to 1 L

1 1„ when said product is run on a computer.