JP3103931B2 - Indoor monitoring device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a room monitoring apparatus using a laser and a pattern projector for generating a grid-like spot beam. More specifically, the present invention relates to a room presence / absence, a position and a person standing indoors. TECHNICAL FIELD The present invention relates to an indoor monitoring device that recognizes a posture state such as a falling state or a falling state quickly and accurately, automatically detects a danger state of a human, and promptly issues a warning.

[0002]

2. Description of the Related Art Heretofore, nursing care for elderly people who enter a nursing home such as a nursing care center or a nursing home has been performed by a caregiver of the institution accompanying the elderly. However, Japan is moving toward an aging society, and the number of elderly living in nursing homes is increasing. On the other hand, there is a shortage of young caregivers who can provide care for the elderly, so that care can be expected to become scarce in the future.

On the other hand, there are quite a few elderly people who are exposed to dangers due to gait disorders or sudden illness in the room, and there is only a call button attached to the room as a means for informing the caregiver of this danger. However,
An elderly person who has been exposed to a situation such as falling indoors may not be able to press the call button, and the caregiver may not be aware of the danger of the elderly person.

[0004] In order to quickly know the danger state of a human being in a room, there is a case where the person is visually monitored using a video camera. However, there is a problem in terms of privacy protection. Methods for automatically detecting the state have been proposed. The principle of this method is as follows. A laser beam in a non-visible light region such as a near-infrared region is formed into a grid-like spot light through fiber grating, and is projected onto a monitoring target. The arrangement of the grid-like spot light group projected on the object is photographed by a video camera that captures an image of the wavelength of the laser light. The spatial position of each spotlight in the room is calculated from the captured image.

[0005] According to the monitoring apparatus of such a system, it is possible to monitor the state of a large object such as a patient, although the details of the object cannot be seen. In Japanese Patent Application Laid-Open No. H5-161613, a fiber grating and a laser are used to monitor the condition of a patient even when the nurse is not present at the patient's side and automatically notify the nurse when an abnormality occurs. An in-room patient monitoring device used has been proposed.

[0006] Since a patient monitoring apparatus in a hospital room using fiber grating and laser does not observe the details of an object, if a patient and a moving object are present in the same room, it may be difficult to identify the patient and the moving object. In order to solve this, an apparatus as described in JP-A-8-150125 has been proposed. In this apparatus, as a method of separating a patient and a moving object, a method of recognizing a moving object to the end as a patient has been proposed. Further, this device is provided with alarm generating means for an abnormal state of the patient.

[0007]

The monitoring apparatus disclosed in Japanese Patent Application Laid-Open No. 8-150125 is not suitable for use in patients whose movement range is limited to a bed and around the bed, such as a severely ill patient after surgery. There is no problem, but there is a problem when it is used for patients who mainly work on the floor, such as elderly people and mentally ill patients. FIG. 9 is a diagram illustrating this problem. FIG. 9a shows the monitoring range of a patient whose movement range is limited to the bed and around the bed. According to the monitoring device of Japanese Patent Application Laid-Open No. 8-150125, there are two types of “floor zone” and “bed zone”.
By providing two zones, the patient and the moving object can be separated in certain cases. For example, if the patient moves from the “floor zone” to the “bed zone”, there is a change spot on the bed, so it can be recognized that the patient is on the bed,
Objects remaining on the floor can be recognized as moving objects.

However, in the method of separating a human and a moving object by dividing into zones, when a human and a moving object coexist in one zone, the two cannot be separated. 9b, 9c and 9d show the case where the monitoring range is only the floor. Let's divide the floor into three zones: “floor 1 zone”, “floor 2 zone”, and “floor 3 zone”. When a person falls from the wheelchair on the floor, as shown in FIG. 9b, if there is one zone between the zone to which the fallen person belongs and the zone to which the wheelchair belongs, the person and the wheelchair can be separated.
However, when the zone to which the person belongs and the zone to which the wheelchair belongs are adjacent as shown in FIG. 9C, or when the person and the wheelchair belong to the same zone as shown in FIG. 9D, the person and the wheelchair cannot be distinguished, You can't tell if a person is sitting on a wheelchair, standing up, or falling on the floor.

[0009]

SUMMARY OF THE INVENTION In order to solve the above-mentioned problems, the present invention is directed to a room in which a person to be monitored, a moving object such as a walking machine or a wheelchair, and a fixed object such as a bed exist. A device that projects spots divided into a three-dimensional lattice pattern, captures an image of the projected image in a predetermined cycle, and monitors the position and posture of the person from changes in the coordinates of each spot. (1) first reference data storage means for storing coordinate data of each spot in a room only of a fixed object and a moving object and no human being, and (2) coordinates of each spot irradiated on a non-human portion of the room. Second reference data storage means for storing data; (3) eight neighboring spot coordinate data storage means for storing coordinate data of eight neighboring spots of each spot stored in the first reference data storage means; ) In the latest cycle The latest spot coordinate data calculating means for calculating the coordinates of each captured spot, (5) the previous spot coordinate data storing means for storing the coordinates of each spot output in the previous cycle, and (6) the latest spot coordinate data A one-cycle moving spot extracting means for extracting a spot moved during the past one cycle from a difference between the output of the calculating means and the contents of the previous spot coordinate data storing means; (7) an output of the latest spot coordinate data calculating means and (2) grouping means for obtaining a moving spot from a difference from the contents of the reference data storage means, and grouping the moving spots into spot groups adjacent to each other based on the contents of the moving spot and the eight neighboring spot coordinate data storage means; (8)
Human group extracting means for extracting a group in which a person exists based on the output of the grouping means and the output of the moving spot extracting means for one cycle; and (9) the output of the grouping means, the output of the human group extracting means, and the latest spot. Second reference data updating means for updating the second reference data based on the output of the coordinate data calculating means; and (10) a human in the room based on the contents of the first reference data storage means and the contents of the second reference data storage means. Non-human non-moving object reference data calculating means for calculating coordinates of a spot irradiated on the object other than the moving object, and (11) non-human non-moving object reference data storage for storing the output of the non-human non-moving object reference data calculating means Means, (12) human position specifying means for specifying a position where a person exists based on the output of the human group extracting means, and (13) output of the latest spot coordinate data calculating means, non-human non-moving object reference data. (14) human position recognition means for recognizing the human posture state based on the contents of the data storage means; Means, and (15) a warning generating means for generating a warning when a person continues a specific state for a preset time or more.

[0010]

FIG. 1 is a flow chart for explaining the operation of the apparatus of the present invention. A spot light divided into a lattice pattern is projected onto an observation target to capture an image. The coordinate data of the spot is calculated based on the captured image of the spot light. The coordinate data of the spot taken in the latest cycle is called the latest spot coordinate data.

Each spot coordinate data in the state where there is no person in the room at the start of operation is called "first reference data". The coordinate data of each spot in the case where it can be assumed that there is no human in the room at this time is referred to as “second reference data”.

FIG. 2 shows an example of eight neighboring spots (A1, A2, A3, A4, A5, A6, A7, A8) with respect to the noted spot T. The difference between the latest spot coordinate data and the spot coordinate data captured in the previous cycle is obtained, and the spot where the coordinate has changed is called a “spot moved for one cycle”. The moving spot for one cycle is a spot irradiated on a moving person or a moving object.

FIG. 3 is an explanatory diagram of grouping. A moving spot as shown in FIG. 3A is obtained from the difference between the latest spot coordinate data and the second reference data. These moving spots mean either moving people or moving objects.

Next, spots adjacent to each other are grouped, and the spots are divided into groups. In the example of FIG.
The spots are group A and group B, as shown in (b).
Into two groups. In the group, the number of moving spots for one cycle is larger than a certain threshold, and
A group having the largest number of moving spots during one cycle is determined to be a human group.

FIG. 4 is an explanatory diagram of a method for determining a human group. A black point is a moving spot obtained by grouping, and a white point is a moving spot for one cycle. The number of moving spots in one cycle belonging to group A is three, and the number of moving spots in one cycle belonging to group B is eight. Therefore, the group B is determined to be a human group, and the group A is determined to be a non-human group. When the human group is extracted, the coordinate data of the spot belonging to the non-human group in the second reference data is updated with the latest spot coordinate data.

Based on the first reference data and the second reference data, a spot position when no human or moving object exists (hereinafter referred to as "non-human non-moving object reference data") is calculated. The coordinates of the spot irradiated on the moving object included in the first reference data and the second reference data are changed to the coordinates of the spot when the moving object disappears (at this time, almost the spot is irradiated on the floor). Convert to coordinates and output.

The position where a person is located is specified from the coordinates of the human group. The difference between the non-human non-moving object reference data and the latest spot coordinate data is calculated, and the three-dimensional information of the human or the moving object, especially height information, is calculated from the movement amount of the spot where the coordinate changes, and the posture of the human is recognized. I do. Based on the indoor map, the position of the person, and the posture of the person, the state of the person, such as “not in the room”, “falling in the room”, or “being on the bed”, is determined.

Prior to operating the apparatus of the present invention, the position and posture of a person considered to be dangerous, and the state duration before a warning is issued, are set. When the state continues for the duration or more, an alarm is automatically generated.

FIG. 5 is an explanatory diagram of an example of detecting an operation in which a person moves out of a room after moving an object. In FIG. 5, five consecutive cycles (n−4) sampled with respect to the time axis T
The image of the cycle... Nth cycle) is shown.

In (a), a rectangle indicates a moving object, and an ellipse indicates a human. (B) shows a group of spots irradiated on the moving object. (C) shows a group of moving spots during one cycle. (D) shows a group. (E) shows a human group.

In the n-4th cycle, there is a moving object in the room. In the second reference data, coordinates of a spot irradiated on the moving object are stored. The number of moving spots during one cycle is zero. Therefore, no human group has been extracted.

At the (n-3) th cycle, a human entered the room. The second reference data is not different from the second reference data in the (n-4) th cycle. A group of moving spots for one cycle appears where a human is, and one group is extracted. This group is extracted as a human group because humans are considered to be in the position of this one group.

In the (n-2) th cycle, a human moved the moving object. Since the human group was specified in the (n-3) th cycle, the coordinates of all the spots constituting the spot group irradiated on the moving object in the (n-3) th cycle of the second reference data are the spots illuminated on the floor. Is updated to the coordinates of In addition, spots irradiated to both humans and moving objects are extracted as a group of moving spots for one cycle,
One group is extracted, and this group is extracted as a human group.

In the (n-1) th cycle, the human proceeded while leaving the moving object. Since the human group was specified in the (n-2) th cycle, the coordinates of all the spots constituting the spot group irradiated on the moving object in the (n-1) th cycle of the second reference data are updated to the coordinates of the current spot. Is done. Since the moving object is abandoned and has not moved, it is extracted as a group of moving spots for one cycle only where a human is present, one group is extracted, and this group is extracted as a human group.

In the n-th cycle, the second reference data remains unchanged. The number of moving spots during one cycle is zero.
Therefore, no human group has been extracted.

From the results of the above (a) to (d), as shown in (e), no human was present at the (n-4) th cycle. In the n-3th cycle, a human entered the room. n-2
In the cycle, humans moved indoors. In the (n-1) th cycle, a human moved indoors. In the nth cycle,
A human has left the room.

[0027]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of an indoor monitoring apparatus according to the present invention will be described. FIG. 6 is an explanatory diagram of the indoor monitoring device according to the present embodiment. The sensor optical system of the image capturing device 1 is a laser projector 1
05, a pattern light projecting element 104, a camera 103, a lens 101, and an infrared transmission band-pass filter 102. In the laser projector 105, an infrared laser beam emitted from a laser diode driven by a laser driving unit 106 is collimated, Output as a laser beam. The laser beam passes through the pattern light emitting element 104 and is divided into a grid spot pattern. In this embodiment, fiber grating is used as the pattern light emitting element.

The projected spot projected image in the room is taken by the camera 103, and in this embodiment, a CCD camera is used. Between the camera 103 and the lens 101, an infrared transmission band-pass filter 102 for transmitting only the laser oscillation wavelength is mounted,
In the case of the present embodiment, an interference filter is used as an infrared transmission band-pass filter. The infrared transmission band-pass filter 102 blocks disturbance light having a wavelength other than light reflected from a spot projected by the laser. The spot projection image is taken into the frame memory 2 from the camera 103, and the data processing device 3 performs the processing shown in the flowchart of FIG. 1 on the spot projection image on the frame memory 2.

FIG. 7 is a view showing a state in which the device of the present invention is installed in a nursing care facility. In room 600, toilet 6
02 and a bed 601. The old man lives alone.
Two sensors 603 and 604 are installed in the room 600. 620 is a service station where a caregiver is waiting. In the service station 620,
An operation touch panel monitor 621 is provided. Control unit (frame memory and data processing device) of this device 6
10 are installed in the ceiling of the service station. The cables 630 connect the sensors 603 and 604 to the control unit 610. The control unit 610 and the operation touch panel monitor 621 are connected by a cable 640.

The contents of the processing of the data processing apparatus will be described with reference to the flowchart of FIG. The spot coordinate data calculated in a state where no person is present in the room is the first reference data. Immediately after the start of operation of the apparatus, the second reference data is the same as the first reference data. The present apparatus confirms that there is no human in the room at the start of operation, and then takes the first reference data only once. The coordinates of the eight neighboring spots are calculated based on the spot coordinate data of the first reference data.

The extraction of the human group is performed using the results of the extraction and grouping of the moving spots for one cycle. 1
The inter-cycle movement spot extraction calculates a difference between the latest spot coordinate data and the previous spot coordinate data, and extracts a spot moved during one cycle. These spots are due to the movement of humans or moving objects in the room.

In the grouping, first, a difference between the second reference data and the latest spot coordinate data is obtained to determine a moving spot. Next, a moving spot group called a group is formed based on eight neighboring spots of these moving spots. Extract.
Lastly, a group including the largest number of movement spots for one cycle among the extracted groups is defined as a human group. If nothing moves during one cycle, the number of moving spots during one cycle is 0, and no human group is extracted. In this case, the human group specified in the previous cycle is set as the human group in the current cycle.

The spots not belonging to the human group in the second reference data are updated with the latest coordinate data. Therefore,
When the human group is extracted, the second reference data is updated.

The first reference data includes a spot irradiated on the moving object at the start. The second reference data also includes a spot irradiated on the moving object in the current cycle. In order to obtain human height information, it is necessary to determine spot coordinates when the spot hits the floor. Therefore, the coordinate data of the first reference data and the second reference data of all spots are compared, and coordinate data corresponding to a low irradiation position is selected, and the collected data is used as non-human non-moving object reference data.

The human condition comprehensive judgment is performed on the basis of human position identification, human posture recognition, human condition judgment, special area data, and installation information. The special area data storage stores bed position information, door position information, toilet position information, and the like. The installation information stores information such as the sensor installation height of the apparatus and the position where the sensor is installed.
At the time of installation of the device according to the invention,
A maintenance person inputs the information using the touch panel monitor 621.

In the human position specification, the center of the human group extracted by the human group extraction is obtained, and the center position is determined based on which special region among several special regions in the special region data is included. Identify where the person is on the floor, bed, or toilet. In this embodiment, the median of the coordinate data of the spots included in the human group is obtained, and this is set as the center of the human group. In the human posture recognition, the three-dimensional position of the point irradiated with each spot is obtained based on the movement amount of the moving spot detected by comparing the latest spot coordinate data with the non-human non-moving object reference data, And three-dimensional human information.

In the human state determination, the human state such as "standing on the toilet", "falling on the floor", "sleeping on the bed" or the like is determined based on the data of the human position identification and the human posture recognition. .

The human state time series data stores the time series data of the human state judgment result. As the warning generation condition, a continuation time condition for generating a warning when a dangerous state assumed by a human state continues is stored in advance. When the condition set in the warning generation condition is satisfied by the contents of the human state time series data, a warning is generated.

[0039]

[Table 1]

Table 1 shows an example until a warning is generated. FIG. 8 shows a flowchart of the operation. In this example, “fall in the room” is a danger state, “stand in the room” is a safe state, and the duration time condition is 5. When a person falls into a danger state, the timer is counted up. If a human enters a safe state before the timer exceeds the duration condition, the timer is reset. Generates a warning if the timer exceeds the duration condition. The above duration condition is input by the caregiver through the touch panel monitor 621 in the service station 620 before monitoring is started.

[0041]

According to the present invention, since it is possible to separate and recognize a person and a moving object by grouping, a person on the floor can be distinguished from other objects on the floor. Therefore, when a person using the wheelchair, such as a nursing care sensor or a resident in a nursing home, leaves the wheelchair, it is possible to accurately detect whether the person is standing or falling.

According to the present invention, since the position of a person in a room can be accurately obtained by grouping and specifying the position of a person, it is possible to accurately detect that a person has entered a specific area. Therefore, when monitoring a person who is in danger of entering a specific area, an alarm can be prevented if an alarm is issued when a person enters the area. In particular, an elderly person with dementia or a psychiatric patient who is monitored by the monitoring device according to the present invention may attempt suicide by jumping out of a window or hitting a head against a wall. According to the present invention, such accidents can be prevented.

Further, since the position and posture of the monitored person are automatically recognized by the calculation using only the coordinates of the spots in the image, the caregiver does not need to always look at the image in the room and the privacy of the human can be protected. It is also possible to centrally monitor persons in multiple rooms at remote service stations.

Further, in the present invention, the data processing for recognizing the position and posture of a human uses only the coordinates of spots in an image, and is much faster than the calculation using the image directly. It is. Therefore, human operation can be completely tracked with a processing capability of about one-chip microcomputer.

[Brief description of the drawings]

FIG. 1 is a flowchart illustrating the operation of the present invention.

FIG. 2 is an explanatory diagram showing an example of eight nearby spots with respect to a spot of interest T;

FIG. 3 is an explanatory diagram of a grouping unit according to the present invention.

FIG. 4 is an explanatory diagram of a human group determination method according to the present invention.

FIG. 5 is an explanatory diagram of an example of detecting an action of a person moving out of a room after moving an object in the embodiment of the present invention.

FIG. 6 is a block diagram of an apparatus according to an embodiment of the present invention.

FIG. 7 is an explanatory view showing a state in which the apparatus according to the embodiment of the present invention is installed.

FIG. 8 is a flowchart until a warning occurs in the embodiment of the present invention.

FIG. 9 is an explanatory diagram of an example of detecting an action of a person getting off a wheelchair according to the effects of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Image capture device 2 Frame memory 3 Data processing device 4 Touch panel monitor 5 Alarm device 101 Lens 102 Infrared transmission type band pass filter 103 Camera 104 Pattern light projecting element 105 Laser projector 106 Laser drive part 600 Room 601 Bed 602 Toilet 603 Sensor 604 Sensor 610 Control unit 620 Service station 621 Touch panel monitor 630 Cable 640 Cable

────────────────────────────────────────────────── ─── Continuing on the front page (72) Koichi Nakagawa, Inventor 1-6-10-10, Tomobuchi-cho, Miyakojima-ku, Osaka-shi, Osaka (72) Shinji Yamada 3-4-17-1602, Tsurumi, Tsurumi-ku, Osaka-shi, Osaka (72) Inventor Junichi Yamaguchi 1-6-6 Moto-Akasaka, Minato-ku, Tokyo Inside Sogo Security Security Co., Ltd. (72) Inventor Kashioka At 1-6-6 Moto-Akasaka, Minato-ku, Tokyo Inside Sogo Security Security Co., Ltd. (56) References JP-A-8-150125 (JP, A) JP-A-3-1299991 (JP, A) JP-A-8-320934 (JP, A) JP-A-1-232497 (JP, A) ( 58) Surveyed fields (Int.Cl. ⁷ , DB name) G08B 23/00-31/00 A61B 5/00 H04N 7/18

Claims (1)

(57) [Claims] 1. A human, a moving object such as a walker or a wheelchair,
A spot divided into a two-dimensional lattice pattern is projected into a room where a fixed object such as a bed is present, and the projected image is captured in a predetermined cycle. A device that monitors the position and posture of the human by measuring a three-dimensional shape,
(1) First reference data storage means for storing coordinate data of each spot where no person is present in the room, and (2) second reference data storage means for storing coordinate data of each spot irradiated on a part other than a person in the room. Reference data storage means; (3) 8-neighbor spot coordinate data storage means for storing coordinate data of eight neighbor spots of each spot stored in the first reference data storage means; and (4) fetched in the latest cycle. (5) a latest spot coordinate data calculating means for calculating the coordinates of each spot, (5) a previous spot coordinate data storing means for storing the coordinates of each spot output in the previous cycle, and (6) a latest spot coordinate data calculating means. Moving spot extracting means for one cycle for extracting a spot moved in the past one cycle from the difference between the output and the contents of the previous spot coordinate data storing means; (7) latest spot coordinates The moving spot is obtained from the difference between the output of the data calculation means and the content of the second reference data storage means, and the moving spot is determined for each adjacent spot group based on the contents of the moving spot and the eight neighboring spot coordinate data storage means. Grouping means for grouping,
(8) an output of the grouping means, a human group extracting means for extracting a group in which a person exists based on an output of the moving spot extracting means for one cycle, and (9) an output of the grouping means and an output of the human group extracting means. Second reference data updating means for updating the second reference data based on the output of the latest spot coordinate data calculating means, and (10) the contents of the first reference data storage means and the contents of the second reference data storage means. A non-human non-moving object reference data calculating means for calculating coordinates of a spot irradiated on a room other than a human and a moving object in a room; and (11) a non-human non-moving object storing an output of the non-human non-moving object reference data calculating means. Reference data storage means, (12) human position specifying means for specifying a position where a person is present based on the output of the human group extracting means, and (13) output of the latest spot coordinate data calculating means, non-human non-moving object reference Human posture recognition means for recognizing a human posture state based on the contents of the data storage means;
4) human state determining means for determining the state of the human based on the output of the human position specifying means and the output of the human posture recognizing means; And a warning generating means for generating an alarm.