JP2000285223A - Fall detector - Google Patents

Abstract

PROBLEM TO BE SOLVED: To detect not only a fallen state but also a falling motion itself with high precision by detecting a fall of a dweller based on changes per time of height information of the dweller obtained by correction of an error to be generated by difference in distance from the dweller to a wall camera. SOLUTION: An image of the dweller 30 is picked up from the side by the wall camera 10 and is picked up from the upper direction by a ceiling camera 11. The image of the dweller 30 is extracted from the picked up video picked up by the cameras 10, 11, the distance from the dweller 30 to the wall camera 10 is calculated from the video picked up by the ceiling camera 11 and the height information of the dweller 30 is calculated from the video picked up by the wall camera 10 by an image processing means 20. The error to be generated by the difference in the distance from the dweller 30 to the wall camera 10 is corrected regarding the height information of the dweller 30 by a height correcting means 21. A judging means 22 detects the falling motion of the dweller 30 based on the changes per time of the height information of the dweller 30 obtained by correction.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a fall detecting device capable of automatically detecting that a resident has fallen.

[0002]

2. Description of the Related Art In order to automatically detect an abnormal condition occurring in a resident requiring care, such as an elderly person, a device using an image processing technique has been conventionally devised. These devices acquire three-dimensional information of a room by image processing and determine an abnormal state of a resident based on the information.

[0003] Some devices of this type are intended to detect a falling motion of a resident. This device calculates three-dimensional information indicating the position where a patient is present by making full use of the principle of triangulation from an image captured by a camera, and detects a state where the patient is lying on the floor surface. If the state continues for a predetermined time from, it is determined that the patient has fallen, and a notification to that effect is given.

[0004]

However, in the fall detecting device having the above-mentioned structure, the light emitting means for projecting the patient's room to obtain three-dimensional information and the projected image are displayed. Since a camera or the like for imaging is required, the apparatus itself becomes large-scale, and there is a problem that a patient in a living room where the apparatus is installed due to the enlargement of the apparatus itself may give a feeling of oppression. Was. Further, the conventional fall detecting device detects that the patient is in a fall state, and has a problem that the fall operation itself cannot be detected.

SUMMARY OF THE INVENTION The present invention has been made in view of the above-mentioned problems, and has as its object to provide a low-cost method capable of accurately detecting not only a falling state but also a falling operation itself. An object of the present invention is to provide a fall detection device.

[0006]

According to the first aspect of the present invention,
A wall camera that images the resident from the side, a ceiling camera that images the resident from above, and distance information from the occupant to the wall camera and height information of the occupant are calculated using the image captured by the camera. Image processing means, and height correction means for correcting an error caused by a difference in the distance between the resident and the wall camera with respect to the resident's height information, and the resident's height information per time obtained by the correction. Determining means for detecting a falling motion of the resident based on the change.

According to a second aspect of the present invention, in the fall detection device according to the first aspect, a change in height information of the occupant is detected as a change in the highest point in an image of the occupant. Is what you do.

[0008]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, a fall detecting device according to an embodiment of the present invention will be described in detail with reference to FIGS. FIG. 1 is a schematic configuration diagram of a fall detection device. FIG.
Is an explanatory diagram showing a process of calculating a perpendicular distance from the highest point to the optical axis of the wall camera based on the image captured by the wall camera, (a) shows the state of the living room,
(B) shows an image captured by the wall camera. FIG. 3 is a schematic diagram of an image captured by a wall camera, (a) is a background image captured when no resident is present, (b) is an input image, and (c) is an input image. 11 is a background difference image extracted by performing a difference process between the image and the background image. 4A and 4B are schematic diagrams showing a resident and the highest point thereof in a background difference image when the resident has fallen, FIG. 4A shows a state before the resident has fallen, and FIG. (C) shows a state in which the resident has completely fallen. 5A and 5B are explanatory diagrams illustrating a process of calculating a distance from a wall camera to a resident based on an image captured by a ceiling camera, where FIG. 5A illustrates a state of a living room, and FIG. 5 shows an image captured by the computer. FIG. 6 is a graph showing the relationship between the height of the highest point of the background difference image and time when a resident falls. FIG. 7 is a graph showing the relationship between the change of the highest point and the time when the resident falls. FIG. 8 is a graph showing the relationship between the height of the highest point of the background difference image and time when the occupant is seated. FIG. 9 is a graph showing the relationship between the change of the highest point and the time when the occupant sits.

The fall detecting device according to the present embodiment is shown in FIG.
As shown in FIG. 1, a wall camera 10 for imaging the occupant 30 from the side and a ceiling camera 11 for imaging the occupant 30 from above.
And an image of the resident 30 is extracted from the video captured by the cameras 10 and 11, the distance from the resident 30 to the wall camera 10 is calculated from the video captured by the ceiling camera 11, and the image is captured by the wall camera 10. Information image processing means 20 for calculating resident height information from the captured video, and resident 30
Height correction means 21 for correcting an error caused by a difference in the distance to the occupant, and determination means 22 for detecting the occupant's overturning operation based on a change in occupant height information per unit time obtained by the correction. Be prepared.

In the present embodiment, the wall camera 10
Is installed on the wall surface of the resident's 30 room, and the ceiling camera is installed on the ceiling of the resident's 30 room so that the resident 30 can be imaged from the side and above. The wall camera 10 and the ceiling camera 11 output video signals of the occupants 30, and the image processing means 20 captures the video. In the image processing means 20, after performing difference processing on the image captured by the wall camera 10 and extracting the occupants, the highest point of the obtained background difference area image is searched, Camera 10
Then, a perpendicular distance dh (t) (see FIG. 2) between a horizontal line passing through the optical axis of the wall camera 10 and the uppermost point on the image captured by is calculated. Furthermore, after the occupants 30 are similarly extracted from the image captured by the ceiling camera 11, the distance d (t) between the wall camera 10 and the occupants is calculated. Also, no matter where the resident 30 is in the room, the height correction means 2 is provided so that the position of the highest point can be compared regardless of the distance between the wall camera 10 and the resident 30.
In 1, a correction is made to the perpendicular distance dh (t) between the horizontal line passing through the optical axis of the wall camera 10 and the uppermost point. The judging means 22 detects the falling action of the occupant 30 based on the corrected height information of the highest point.

The image processing means 20 converts an image signal captured by the cameras 10 and 11 into a digital signal.
A converter, an image memory for storing the converted video signal,
Difference means for calculating a difference between a newly captured input image (see FIG. 3B) and a background image (see FIG. 3A) captured when there is no resident, and the obtained background difference It is provided with a difference image memory for storing an image (see FIG. 3C).

The image processing means 20 having the above-described configuration continuously performs difference processing for each input image captured by the wall camera 10. Then, after performing the binarization process and the labeling process on the background difference image, in the background difference image (see FIG. 3C) obtained by the image processing means 20, the highest point of the region of the background difference region image is obtained. Detection is performed. Further, a perpendicular distance dh (t) between the extracted highest point and a horizontal line passing through the optical axis of the wall camera 10 is calculated.

The image processing means 20 includes the ceiling camera 1
1, the captured image is captured, a background difference image is generated by the same procedure as described above, and the binarization process and the labeling process are performed. Further, as shown in FIG. 5, the distance d (t) is calculated from the position coordinates of the area (resident 30) of the background difference image obtained by the ceiling camera 11 and the position coordinates of the wall camera 10.

The height correcting means 21 corrects the height of the uppermost point using the distance d (t) between the resident 30 and the wall camera 10. In a camera-captured image representing a three-dimensional space as a plane, the height dh (t) increases as the distance d (t) from the wall camera decreases and decreases as the distance d (t) decreases. Have. Therefore, in order to correct this distortion, the distance d is added to the height dh (t).
Height dh 'corrected for distortion by multiplying by (t)
(T) = a × dh (t) × d (t) is used as the height of the corrected highest point. Where a is the focal length,
This is a constant determined by the size of the CCD camera, the number of pixels, and the like. This makes it possible to accurately capture the height change of the highest point regardless of the position of the human body 30 in the living room, regardless of the distance from the wall camera 10 to the human body 30.

The judging means 22 accumulates the height dh '(t) of the highest point at each time corrected by the height correcting means 21, monitors a change in the height of the highest point, and turns over the resident. Is to be detected. Judgment of a fall is based on the fact that the movement of the resident's head, that is, the highest point per unit time is extremely large at the time of a fall. The height of the highest point is used as a value that can easily approximate the height of the resident's head. This is because, as shown in FIG. 4, when the resident is walking (see FIG. 4 (a)) or falling (see FIG. 4 (b)), the height of the highest point and the resident The height of the person's head is the same, and the height of the top point and the height of the head are almost the same even after a fall.

In the fall detecting device according to the present embodiment,
As shown in FIG. 6, the relationship between the height of the highest point and time is monitored, and as shown in FIG. 7, the degree of change in the position of the highest point at each time is calculated, and the amount of change in the highest point is calculated. Is smaller than the threshold th, it is determined that the resident has fallen.
In the present embodiment, the change of the highest point is calculated as a positive value when the position of the highest point is higher, and is calculated as a negative value when the position of the highest point is lower. Also, as shown in FIGS. 8 and 9, the height of the highest point gradually changes during the occupant's sitting operation, and does not change so as to exceed the threshold value th. It is not to be done.

In the fall detecting device according to the present embodiment,
Since it does not have a conventional light projecting means, it can be manufactured compactly and at low cost, and the occupant's movement clearly appears from images taken by the wall camera 10 and the ceiling camera 11. It is possible to detect the characteristics of the resident's falling motion based on the amount of change in the resident's head that is easy to detect. It becomes possible to detect the fall of the resident well.

In the first and second embodiments, a background difference for calculating a difference between a previously captured background image and an input image is used as difference processing.
An inter-frame difference for performing a difference process between captured images may be used. Also, in the present embodiment, attention is paid to the height of the highest point of the background difference image, because the change in height of the resident can be easily grasped. However, it is needless to say that the present invention is not limited to this as long as the change in height information of the resident 30 per unit time can be captured.

[0019]

As described above, according to the first aspect of the present invention, a wall camera that images a resident from the side, a ceiling camera that images a resident from above, and an image that is captured by the camera Image processing means for calculating the distance from the resident to the wall camera and the resident's height information using the video, and correcting the error caused by the difference in the resident's height information from the resident to the wall camera. Height correction means;
A determination means for detecting a falling motion of the resident based on a change in height information of the occupant per time obtained by the correction, so that a light emitting means other than a wall camera for photographing the living room is provided. The structure is simple because it is not required, and the resident falls with only a simple process of calculating the resident's height information that causes a characteristic change when falling without depending on the distance between the resident and the wall camera. Since the operation can be detected with high accuracy, it is possible to provide a low-cost fall detection device capable of accurately detecting not only a fall state but also a fall operation itself.

According to the second aspect of the present invention, the first aspect is provided.
In the fall detection device described above, a change in height information of the occupant is detected as a change in the highest point in the image of the occupant, so that the occupant's movement clearly appears and the occupant can be easily detected. Since it is possible to detect the feature of the resident's falling motion using the head of the occupant, it is possible to easily and accurately detect the resident's falling.

[Brief description of the drawings]

FIG. 1 is a schematic configuration diagram of a fall detection device.

FIG. 2 is an explanatory diagram showing a process of calculating a perpendicular distance from an uppermost point to an optical axis of the wall camera based on an image captured by the wall camera, wherein (a) shows a state of a living room; b) shows an image taken by the wall camera.

3A and 3B are schematic diagrams of images captured by a wall camera, FIG. 3A is a background image captured when no resident is present, FIG. 3B is an input image, and FIG. It is a background difference image extracted by performing difference processing between an image and a background image.

FIG. 4 is a schematic diagram showing a resident and its highest point in a background difference image when the resident falls, (a) shows a state before the resident falls, and (b) shows a state before the resident falls. FIG. 3C shows a state in which the occupant has fallen completely, and FIG.

FIG. 5 is an explanatory diagram illustrating a process of calculating a distance from a wall camera to a resident based on an image captured by a ceiling camera, where (a) illustrates a state of a living room;
(B) shows an image captured by the ceiling camera.

FIG. 6 is a graph showing the relationship between the height of the highest point of the background difference image and time when a resident falls.

FIG. 7 is a graph showing the relationship between the change of the highest point and time when a resident falls.

FIG. 8 is a graph showing the relationship between the height of the highest point of the background difference image and time when a resident is seated.

FIG. 9 is a graph showing the relationship between the change of the highest point and time when a resident sits down.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 Wall camera 11 Ceiling camera 20 Image processing means 21 Height correction means 22 Judgment means 30 Resident

[Procedure amendment]

[Submission date] December 16, 1999 (1999.12.
16)

[Procedure amendment 1]

[Document name to be amended] Statement

[Correction target item name] 0009

[Correction method] Change

[Correction contents]

The fall detecting device according to the present embodiment is shown in FIG.
As shown in FIG. 1, a wall camera 10 for imaging the occupant 30 from the side and a ceiling camera 11 for imaging the occupant 30 from above.
And an image of the resident 30 is extracted from the video captured by the cameras 10 and 11, the distance from the resident 30 to the wall camera 10 is calculated from the video captured by the ceiling camera 11, and the image is captured by the wall camera 10. and images processing unit 20 you calculated height information of the resident from the video, height correction means for correcting the errors caused by the difference in distance from the resident 30 to the wall surface camera 10 the height information of the resident 21 and judgment means 22 for detecting the occupant's overturning operation based on a change in occupant height information per unit time obtained by the correction.

[Procedure amendment 2]

[Document name to be amended] Statement

[Correction target item name] 0014

[Correction method] Change

[Correction contents]

The height correcting means 21 corrects the height of the uppermost point using the distance d (t) between the resident 30 and the wall camera 10. In a camera-captured image representing a three-dimensional space as a plane, the height dh (t) increases as the distance d (t) from the wall camera decreases and decreases as the distance d (t) decreases. Have. Therefore, in order to correct this distortion, the distance d is added to the height dh (t).
Height dh 'corrected for distortion by multiplying by (t)
(T) = a × dh (t) × d (t) is used as the height of the corrected highest point. Where a is the focal length,
This is a constant determined by the size of the CCD light receiving surface , the number of pixels, and the like. This makes it possible to accurately capture the height change of the highest point regardless of the position of the human body 30 in the living room, regardless of the distance from the wall camera 10 to the human body 30.

[Procedure amendment 3]

[Document name to be amended] Drawing

[Correction target item name] Fig. 6

[Correction method] Change

[Correction contents]

FIG. 6

[Procedure amendment 4]

[Document name to be amended] Drawing

[Correction target item name] Fig. 8

[Correction method] Change

[Correction contents]

FIG. 8

 ────────────────────────────────────────────────── ─── Continuing from the front page (72) Inventor Kenichi Hagio 1048 Kazuma Kadoma, Kadoma-shi, Osaka F-term in Matsushita Electric Works, Ltd. QQ28 5B057 AA07 AA19 BA02 BA19 CA12 CA16 DA07 DA15 DB02 DC02 DC33 5C054 CC03 FC01 FC12 FC13 FC15 GB12 HA12 5C087 AA02 DD03 DD49 EE05 EE07 EE14 EE18 FF04 GG02 GG35

Claims (2)

[Claims] 1. A wall camera that images a resident from the side, a ceiling camera that images a resident from above, a distance from the resident to the wall camera using an image captured by the camera, and a distance of the resident. Image processing means for calculating height information, height correction means for correcting an error caused by a difference in the distance between the resident and the wall camera with respect to the resident's height information, and the resident's height obtained by the correction A fall detecting device for detecting a fall operation of a resident based on a change per unit time of information. 2. The fall detecting device according to claim 1, wherein a change in height information of the occupant is detected as a change in the highest point in an image of the occupant.