JP2018082745A - Posture determination device and notification system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a posture determination device by which to accurately determine a posture of a patient, etc. even in such a state that a quilt is used on a bed, and a notification system by which to notify in such a case where there is a possible risk.SOLUTION: A posture determination device 1 acquires a determination image formed with a depth value by determination image acquisition means 10 to eliminate an influence of a quilt, forms divided images by dividing the determination image into images at each constant distance using the depth value in posture determination means 20, and performs posture determination on an object person S using the divided images.SELECTED DRAWING: Figure 8

Description

  The present invention relates to a posture determination device that determines the posture of a subject on a bed and a notification system that reports when there is a risk of falling from the bed by determining the posture of the subject on the bed.

  Various problems related to such a fall accident are known because an accident in which a patient or an elderly person falls from a bed has become a problem. As an example, a sensor that detects the load is placed beside the bed, and when the load is detected by the sensor, a device that reports that a patient has fallen or a sensor is placed on the bed to detect the load. There is known an apparatus that reports that a patient or the like has fallen when it disappears.

  Further, as a technology developed by the inventors of the present application, a system is known that acquires posture information by a sensor attached to a patient in Patent Document 1 and calculates the risk of falling based on the posture information. .

  However, an apparatus that arranges a load sensor beside the bed has many drawbacks such as many false alarms and it is difficult to prevent a fall because it reports after the load is detected. In addition, a device in which a load sensor is arranged on a bed is prone to malfunction due to a difference in body weight, and has problems such as malfunction of the sensor due to sweat or moisture, and many failures. Moreover, the method of attaching a sensor to the patient himself / herself in Patent Document 1 feels uncomfortable with the sensor attached by the patient himself, cannot remove the sensor without permission, or cannot attach the sensor to an appropriate position due to the influence of surgery or the like. There are difficulties such as.

  Therefore, as disclosed in Patent Document 2, an apparatus that acquires an image of a patient on a bed and estimates the posture based on the image is known. The inventor of the present application has also acquired a patient image on the bed and conducted a study to determine the posture based on the image, and the contents thereof have already been published as Non-Patent Document 1.

Japanese Patent Laying-Open No. 2015-103042 JP 2014-236896 A

"Posture estimation on inpatient's bed using vision sensor", Proc. 24th Annual Conference of the Society of Instrument and Control Engineers China Chapter, 2015, p. 150-151

  The posture determination according to Non-Patent Document 1 can be determined with high accuracy when determining the posture of the patient in a situation where the comforter is not used on the bed, but if the posture of the patient is determined in the situation where the comforter is used, A problem has been found that the accuracy of posture determination is significantly reduced by the influence. Assuming that the posture determination device is used at an actual site, the patient on the bed usually uses a comforter. Therefore, it is very important to solve the problem of the accuracy of posture determination due to the comforter.

  Therefore, the present invention provides a posture determination device that can accurately determine the posture of a patient or the like even when a comforter is used on a bed, and a notification system that can report when there is a risk of danger. About.

  In order to solve the above-described problem, the posture determination apparatus according to the present invention divides the determination image into images at fixed distances using determination image acquisition means for acquiring a determination image formed by depth values and the depth value. Posture determination means for performing posture determination of the subject person based on the segmented image.

  In addition, the posture determination unit includes a moving object detection unit that detects a moving object image having a motion from the segmented image, and a combined moving object image forming unit that combines the moving object images of the different segment images to form a combined moving image. And the posture of the subject is determined from the synthesized moving body image.

  In addition, the posture determination unit includes a template matching unit that determines a posture by performing template matching using a template image for the synthesized moving body image.

  In addition, the notification system of the present invention includes a determination image acquisition unit that acquires a determination image formed by a depth value, and a classification image obtained by dividing the determination image into images of a certain distance using the depth value. Posture determination means for performing posture determination, notification determination means for determining whether or not to report based on a determination result by the posture determination means, and notification means for performing notification based on a determination result of the notification determination means It is characterized by providing.

  The posture determination apparatus of the present invention acquires a determination image formed by depth values, and performs posture determination based on segmented images divided into images at fixed distances using the depth values. In a determination image in which an object that interferes with the determination of the posture of the subject, such as a comforter, is present, posture determination can be performed based on a segmented image that has no obstruction.

  Specifically, since moving body images are detected from each segmented image and the posture is determined based on a synthesized moving body image obtained by synthesizing moving body images of different segmented images, obstructive objects such as comforters are detected as moving body images. Even so, the posture of the subject can be determined from the synthesized moving body image until the comforter is detected.

  Specifically, since posture determination is performed by template matching using a template image, posture determination can be performed with high accuracy and at high speed.

  In addition, the notification system of the present invention acquires a determination image formed by the depth value, and performs posture determination based on the segmented image divided into images at fixed distances using the depth value. In a determination image in which an object that interferes with the determination of the posture of the subject, such as a comforter, is present, posture determination can be performed based on a segmented image that does not interfere. In addition, the notification system of the present invention determines whether or not to report based on the determination result of the posture determination, and performs notification based on the determination result. For example, many determinations such as hospitals and care facilities Suitable for use in facilities where the subject is located.

It is a block diagram which shows the hardware constitutions of the attitude | position determination apparatus of Embodiment 1. It is a block diagram which shows the function structure of the attitude | position determination apparatus of Embodiment 1. It is the figure which showed the usage example of the attitude | position determination apparatus of Embodiment 1. FIG. It is the figure which showed the usage example of the attitude | position determination apparatus of a comparative example. (A) is an example of a depth map in the posture determination apparatus of the comparative example, and (B) is an example of an image of the subject detected by the background difference method. (A)-(D) are the specific examples of the attitude | position determination method in the attitude | position determination apparatus of a comparative example. It is an example of the image of the subject detected when the comforter is used in the posture determination apparatus of the comparative example. It is a conceptual diagram of the division | segmentation in the attitude | position determination apparatus of Embodiment 1. FIG. It is an example of the moving body image in the attitude | position P2 detected for every division in the attitude | position determination apparatus of Embodiment 1. FIG. 6 is a flowchart of a combined moving image forming process and a posture determining process performed by a combined moving object image forming unit and a template matching unit in the posture determining apparatus according to the first embodiment. It is an image figure of attitude | position determination in the attitude | position determination apparatus of Embodiment 1. FIG. It is the conditions of the verification environment in verification using the attitude | position determination apparatus of Embodiment 1. FIG. It is a verification result in the posture determination apparatus of Embodiment 1 and the posture determination apparatus of a comparative example, (A) is a verification result without a comforter, and (B) is a verification result with a comforter. It is the figure which showed the usage example of the notification system of Embodiment 2. FIG.

  Hereinafter, specific examples of the present invention will be described in detail with reference to the drawings. However, the embodiment shown below is an example for embodying the technical idea of the present invention, and is not intended to specify the present invention to this embodiment. Other embodiments within the scope are equally applicable.

[Embodiment 1]
First, with reference to FIG. 1 and FIG. 2, a basic configuration of the posture determination apparatus 1 of the present embodiment will be described. FIG. 1 is a block diagram illustrating a hardware configuration of the posture determination device 1, and FIG. 2 is a block diagram illustrating a functional configuration of the posture determination device.

  As shown in FIG. 1, the posture determination apparatus 1 includes an input unit 2, a control unit 3, a storage unit 4, an output unit 5, and a communication unit 6.

  The input unit 2 acquires an image formed based on depth value information for a range including a subject whose posture is determined by the posture determination device 1, and is called a so-called distance image sensor. In the present embodiment, specifically, the input unit 2 acquires a camera 2a for acquiring an image including the target person on the bed and depth information regarding the depth to the subject in the image captured by the camera 2a. And a depth sensor 2b. As a distance image sensor having such a configuration, KINET (registered trademark) manufactured by Microsoft Corporation is known.

  In the present embodiment, the image formed by the depth value is acquired using this KINECT, but other than this, it is needless to say. In addition, KINECT is provided with a microphone for acquiring sound information. Therefore, this microphone may be used as the input unit 2 to acquire sound information. When sound information can also be used for posture determination, the input unit 2 can include a microphone as well as KINECT. In addition to the camera 2a and the depth sensor 2b, the input unit 2 inputs information to the posture determination device 1, for example, the name of a person who is a subject of posture determination, although not shown. For example, an input device such as a keyboard or a touch panel may be included.

The control unit 3 includes a CPU, a RAM, a ROM, and the like, and executes various processes based on stored programs such as a program for performing posture determination described later. The storage unit 4 is a storage device such as a DRAM or HDD, and temporarily stores image data acquired by the input unit 2.
Various data and the like are stored. The output unit 5 includes a display device, a speaker, and the like, and outputs an attitude determination result and the like using video and sound.

  The communication unit 6 communicates with other devices such as a mobile terminal via the Internet or a LAN. Through the communication unit 6, the posture determination device 1 can output the posture determination result and the like to other devices.

  As shown in FIG. 2, the posture determination apparatus 1 includes a determination image acquisition unit 10, a posture determination unit 20, and a determination result output unit 30. The determination image acquisition unit 10 is mainly configured by the input unit 2 and includes a target person who performs posture determination and acquires a determination image formed by a depth value.

  In addition, the posture determination unit 20 is mainly configured by the control unit 3 and the storage unit 4, and uses the depth value included in the determination image for the determination image acquired via the determination image acquisition unit 10. Determine the posture. In particular, the posture determination means 20 according to the present embodiment is characterized in that the determination image is divided into images for every predetermined distance using the depth value, and the posture of the subject is determined using this divided image.

  As an example for realizing such posture determination, the posture determination unit 20 further includes a determination image classification unit 21, a moving body detection unit 22, a combined moving body image formation unit 23, and a template matching unit 24. .

  Although details will be described later along with a specific example, the determination image classification unit 21 divides the determination image acquired via the determination image acquisition unit 10 into an image for each predetermined distance using the included depth value. create. Then, the moving object detection unit 22 detects a moving object image having a movement from the segmented image. Then, the synthesized moving body image forming unit 23 synthesizes moving body images of different segment images to form a synthesized moving body image. Then, the template matching unit 24 uses the synthesized moving body image to perform matching with the template of each posture and determines the matching posture as the posture of the subject person.

  In other words, the posture determination means 20 of the present embodiment creates a segmented image for each predetermined distance in order to eliminate the influence of the comforter, and sequentially synthesizes the moving body images extracted from the segmented images to form a synthesized moving body image. By performing template matching, the posture is determined based on the synthesized moving body image at a distance until the comforter appears.

  The determination result output unit 30 is mainly configured by the output unit 5 and outputs the posture result of the subject determined by the posture determination unit 20 using a monitor or the like. As for the output of the posture result, for example, when the posture of raising the upper body is determined, it is displayed in text or output in voice so that it can be understood that it is “the posture of raising the upper body” . Further, the determination result output means 30 may output the determination result only when it detects a posture in which the subject is highly likely to fall from the bed.

  Next, the posture determination apparatus 1 having such a configuration will be described based on an actual usage example. FIG. 3 is a diagram illustrating a usage example of the posture determination apparatus 1. The posture determination apparatus 1 determines the posture of the subject S who uses the bed B on the bed B using the image acquired by the determination image acquisition unit 10.

  As described above, KINET (registered trademark) manufactured by Microsoft Corporation was used as the distance image sensor of the input unit 2 constituting the determination image acquisition unit 10. Since the input unit 2 needs to photograph the subject S, the input unit 2 is installed in the vicinity of the subject S. However, the hardware constituting the posture determination unit 20 and the determination result output unit 30 is not limited to the bed. It is usually arranged in a room different from the room where B is installed. For example, when it is assumed to be used in a hospital, the input unit 2 is installed on the bed of the inpatient, and other configurations are installed in a place such as a nurse station.

  Note that the posture determination device 1 can also be arranged in the room in which the bed B is installed with respect to the hardware constituting the posture determination means 20 and the determination result output 30. In this case, since it is necessary to notify that the subject S is in a dangerous state by the determination result output means 30, the determination result output means 30 is preferably output by sound such as an alarm sound.

  The installation location of the input unit 2 may be installed on the ceiling of a room, for example, but in this embodiment, as shown in FIG. Is directed toward the subject S's feet. This is for reducing the psychological burden of the subject S that the input unit 2 enters the view and is always monitored by being installed on the head side of the bed B that is difficult to enter the view of the subject S.

  The data of the determination image acquired by the determination image acquisition unit 10 is sent to the attitude determination unit 20 of the attitude determination device 1. Then, based on the determination image acquired by the determination image acquisition unit 10, the posture of the subject S is determined by the posture determination unit 20.

  Here, before describing a specific example of posture determination in the posture determination means 20 in the present embodiment, the specific contents of posture determination previously disclosed in Non-Patent Document 1 by the inventor of the present application will be first described as a comparative example. Do.

[Comparative example]
FIG. 4 is a diagram illustrating a usage example of the posture determination apparatus 1A as a comparative example. The posture determination device 1A of the comparative example has the same hardware configuration as the posture determination device 1 of the present embodiment shown in FIG. In addition, the posture determination device 1A includes a determination image acquisition unit 10A similar to the posture determination device 1 and a determination result output unit 30A. The difference between the posture determination device 1A and the posture determination device 1 is a posture determination unit 20A.

  Similar to the posture determination device 1, the posture determination device 1A first acquires a determination image including a depth value by the determination image acquisition unit 10A made of KINET (registered trademark). Note that this KINECT has a function of acquiring a depth distance in a 512 × 424 [pixel] plane.

  The determination image is obtained by converting depth data from 0.5 to 8.0 (m) into a two-dimensional image as a depth map (Depth map) that is grayscaled at a resolution of 256. FIG. 5A shows an example of the depth map. In addition, infrared rays are used for obtaining the depth map. By using infrared rays, it is possible to read three-dimensional information in a plane even in a dark room at night when there are fewer medical workers and the like, and the privacy of the subject can be protected because the individual's face cannot be identified.

  The subject S is detected as a moving object by the moving object detecting means 21A of the posture determining means 20A using the determination image made up of the depth map. The background subtraction method is used as a specific method of moving object detection. In the background subtraction method, an environment in which no subject S is present is acquired as a background image, and what is not included in the background image is extracted by comparison with the background image. Detect as a moving object.

  FIG. 5B shows an example of an image of the subject S detected as a moving object by the background difference method. Only what was not in the background image is shown in gray in FIG. 5B, and the gray portion is the subject S. In FIG. 5B, the subject is shown in gray, but the device actually used by the inventor is shown in red.

  In addition, when a person other than the target person S moves in the back of the bed B (back side of the target person S), the movement is also detected as a moving object, and erroneous determination occurs. Therefore, using the depth value of the depth map, The moving object detection range is up to 2 m, and distances of 2 m or more are excluded from detection.

  Next, the posture of the moving body image detected by the moving body detection unit 21A is determined by the template matching unit 22A. When performing posture determination, a posture that becomes a point is determined as prior information, and a template is created based on image data of the posture. The template data is stored in a storage unit or the like in the hardware configuration of the posture determination apparatus 1A.

  In the comparative example, three postures were defined as the postures to be points. In FIG. 4, the three postures are a posture indicated as P1: a sleeping posture, a posture indicated as P2, a posture that caused a state, and a posture indicated as P3: a posture sitting on a bed edge. Note that these postures P1 to P3 are also used as the same posture in this embodiment described later.

  In addition, in order to improve the accuracy of posture determination and speed up the determination, a process diagram of posture transition is defined as prior information, and posture transition is also considered. For example, when the current posture is the posture P1, there is no possibility of sudden transition from the posture P1 to the posture P3. In this case, the posture P3 is not set as a posture determination target.

As a specific posture determination method, template matching based on averageHash (hereinafter referred to as “aHash” as appropriate) is used. As a procedure,
1) First, an image of a moving object is detected. A specific example detected is shown in FIG.
2) Next, this image is converted to gray scale. A specific example is shown in FIG. 6A and 6B are the same image here, but the moving body in FIG. 6A is detected as a red image in the actual apparatus of this embodiment as described above. ing. Therefore, FIG. 6B is actually a red image converted.
3) Next, the size of the grayscale image is reduced in order to speed up the processing. This time it has been reduced to a size of 16 × 16.
4) Next, the average value (average_pixel) of each pixel color (shading: pixel_value) of the reduced image is calculated by the equation (1).
5) Next, in each pixel of the reduced image, the bit value of each pixel is set using Expression (2).
6) Next, a 16 × 16 = 256-bit bit string is generated and the image is hashed. A specific example of this bit data is shown in FIG.
7) Next, each hash value of the template images of the postures P1 to P3 generated in advance is shifted by 1 bit and compared. An image of bit comparison is shown in FIG.
8) Then, the minimum value of the bit error at the time of comparison is determined as the matching degree of the image data, and the posture of the template having the highest matching degree is determined as the current posture.

  By such a procedure, the template matching unit 22A of the posture determination unit 20A determines to which template the posture of the target person S is close. And attitude | position determination apparatus 1A outputs a determination result by the determination result output means 30A based on this determination result.

  The posture determination apparatus 1A of the comparative example as described above was able to accurately determine the posture of the subject S when the posture determination was performed on the bed B without a comforter. On the other hand, in the posture determination apparatus 1A, the accuracy of the posture determination of the subject S is significantly reduced in the state where the comforter is used on the bed.

  This is because when the comforter is used on the bed, the comforter also moves as the subject S moves, so that the comforter is detected as a moving object together with the subject S. For this reason, the posture determination apparatus 1A determines that the posture is different from the original posture of the subject S.

  For example, when the subject S is in the posture 2 state while using the comforter C, the moving object as shown in FIG. 7 is detected. For this reason, in the posture determination apparatus 1A, this posture is originally determined as the posture P2, but has been determined as the posture P3 due to the influence of the comforter.

[This embodiment]
Therefore, the posture determination device 1 of the present embodiment can perform posture determination in consideration of comforters. Specifically, the following procedure is performed.

  First, with respect to the determination image acquired by the determination image acquisition unit 10, a determination image classification unit 21 forms a divided image into images of a certain distance. Specifically, using the depth value included in the determination image, the image is divided every 0.1 m to form a divided image for each division. FIG. 8 shows a conceptual diagram of the division so that it is easy to image the divided images divided every predetermined distance.

  As shown in FIG. 8, the image is divided from the input unit 2 toward the feet of the subject S at regular intervals, and an image is formed for each of the sections 1 to n. In addition, in this embodiment, although it classifies every 0.1 m, since the subject S does not exist in the position very close to the input part 2, as shown in FIG. You may start sorting from. Moreover, the interval of the division shown using the broken line in FIG. 8 is only described for description, and does not necessarily indicate the 0.1 m interval in the present embodiment. Further, the current range of 0.1 m is merely an example, and the range of the division can be set shorter or longer.

  With respect to the divided images divided by the determination image dividing means 21 in this way, the moving object detection means 22 detects a moving object image for each divided image. A specific method of detecting the moving body image is the same as that of the posture determination apparatus 1A of the comparative example, and can be performed using the background difference method. Specifically, first, an environment without the subject S and the comforter is acquired as a background image, and the movement of the subject S and the comforter is detected as a moving object image using the background image. FIG. 9 shows the moving body image in each section detected in this way. In addition, the moving body image for every division shown in FIG. 9 is based on the determination image when the subject S is in the state of the posture P2.

  Although the divisions 1 to 12 are shown in FIG. 9, the division 1 side is the input unit 2 side, and the division 12 is the division farthest from the input unit 2 (the foot side of the subject S). As can be seen from FIG. 9, in Sections 1 to 5, a particularly conspicuous moving object cannot be detected, but in Section 6 it can be seen that the back of the subject S is detected. In section 7, it can be seen that the head, shoulders, and body of the subject S are detected. In section 8, it can be seen that not only a part of the subject S but also the comforter C is detected.

  When the comforter C is used on the bed B, the comforter C moves with the movement of the subject S, and the comforter is usually located on the back side (the side away from the input unit 2) from the position of the subject S. become. For this reason, also in the segmented image, the subject S first appears as a moving object on the near side (side closer to the input unit 2), and this can also be understood from FIG.

  Next, the moving body image for each segment image detected by the moving body detecting unit 22 is synthesized by the synthesized moving body image forming unit 23, and the posture matching is performed by the template matching unit 24 using the synthesized moving body image. That is, the posture determination means 20 performs the posture determination of the subject person S with a composite moving body image using the moving body image until the comforter appears.

  Specific processing performed by the synthesized moving body image forming unit 23 and the template matching unit 24 will be described with reference to FIG. FIG. 10 is a flowchart of the combined moving body image forming process and the posture determination process performed by the combined moving body image forming unit 23 and the template matching unit 24.

  First, it is determined whether or not the amount of change in the detected moving object image is greater than or equal to a certain amount from the near-side segment image to be segment 1 (step 1: S1). If the amount of change is not equal to or greater than a certain amount (NO in S1), this process is repeated to determine a segment until a moving object whose amount of change is equal to or greater than a certain amount is detected. This process is for eliminating fine moving objects that are not related to the subject S.

  If the amount of change is equal to or greater than a certain amount (YES in S1), then template matching is performed for the moving object image (step 2: S2). Specifically, this template matching is performed using template matching based on aHash, as in the posture determination apparatus 1A of the comparative example, and detailed description thereof is omitted.

  Then, it is determined whether or not the difference value based on the matching result in S2 satisfies the posture threshold value of any one of postures P1 to P3 (step 3: S3). When the difference value is less than the threshold value of any posture (for example, the difference value is 10 or less) (NO in S3), the moving object image of the next section is synthesized (step 4: S4). Then, template matching is performed on the synthesized moving object image in the same manner as S2 (step 5: S5).

  Then, the difference value based on the matching result in S5 is compared with the difference value based on the matching result in S2, and it is determined whether or not the difference value has been improved (step 6: S6). If the difference value is improved (if the difference value in S5 is lower than the difference value in S2) (YES in S6), the process returns to S3 to determine which posture threshold is satisfied. Judgment is made. If YES in S3, the posture is determined as the posture of the subject (step 8: S8).

  If the difference value is not improved in S6 (if the difference value in S5 is higher than the difference value in S2) (NO in S6), it is determined that the posture is unknown (step 7: S7). Returning to S1, the processing of S1 is performed on the moving object image of the next section.

  The posture determination of the subject person S is performed by performing the processing as described above. A specific image of posture determination will be described using the image of FIG. FIG. 11 shows the moving body of the subject S and the moving body of the comforter detected in four consecutive sections (sections a to d). Then, by performing the processing of FIG. 10 from the section a, the determination result is obtained as the posture P2 in this case by the synthesized moving body image indicated by the section (a + b). That is, the posture determination can be performed with the combined moving image of the section (a + b) until the combined moving image of the section (a + b + c + d) that is affected by the moving body of the comforter appearing in the section d.

  As described above, the posture determination apparatus 1 according to the present embodiment acquires a determination image formed by the depth value by the determination image acquisition unit 10 in order to eliminate the influence of the comforter, and the posture determination unit 20 sets the depth value. The determination image is divided into images for each predetermined distance to form a divided image, and the posture of the subject person S is determined using the divided image. In particular, the posture determination unit 20 detects a moving body image having a motion from the segmented image by the moving body detection unit 22, and forms a synthesized moving body image by combining the moving body images of different segmented images by the synthesized moving body image forming unit 23. The posture determination of the subject person S is performed using the image.

[Verification]
Next, the verification result about the attitude | position determination apparatus 1 of embodiment is demonstrated. Specifically, comparison verification between the posture determination device 1 of the present embodiment and the posture determination device 1A of the comparative example with or without a comforter was performed on five subjects. This comparative verification was performed on the assumption that the subject moved in the order of posture 1 (posture sleeping on the bed), posture 2 (posture of getting up on the bed), and posture 3 (posture sitting on the edge of the bed). did. The verification environment in this verification is shown in FIG.

  And the verification result implemented by this verification condition and verification environment is shown in FIG. FIG. 13A shows the result of verification under the condition without comforter, and FIG. 13B shows the result of verification under the condition with comforter. When there is no comforter as shown in FIG. 13 (A), the correct answer rate was 97.7% on average in both the posture determination device 1A of the comparative example and the posture determination device 1 of the present embodiment. That is, it can be seen that the posture determination apparatus 1 of the present embodiment can perform the same determination as that of the comparative example even in a situation where there is no comforter.

  On the other hand, when the comforter is present as shown in FIG. 13B, the correct answer rate is significantly reduced to 65% in the posture determination device 1A of the comparative example, whereas in the posture determination device 1 of the present embodiment, the comforter is The accuracy rate was the same as when there was no. That is, it can be seen that the posture determination device 1 of the present embodiment can perform the same determination as in the situation without the comforter even in the situation where the comforter is present, and the influence of the comforter can be eliminated. Therefore, only the body part of the target person S can be extracted by acquiring the determination image formed by the depth value and forming a segmented image by dividing the determination image into images of a certain distance using the depth value. As a result, the influence of the comforter is eliminated.

  In addition, in the posture determination apparatus 1 of this embodiment, although the influence of the comforter is eliminated, the correct rate of posture 2 remains 93.3% as shown in FIG. The reason for this is considered to be that the position when getting up differs depending on the subject due to the difference in sitting height of the subject. In other words, the position of the subject at the time of getting up is different, resulting in a difference in the size of the subject shown in the input unit 2, which is considered to be an erroneous determination when determining the posture. In addition, when the subject is severely stooped, there may be a segmented image in which only the back part of the back of the cat is reflected in the segmented image segmented every 0.1 m, and the back was misrecognized as the subject's head. As a result, posture 2 may be erroneously determined as posture 1. As these countermeasures, it is possible to prepare a plurality of template sets according to the characteristics of the patient, or to distinguish between the head and the back from the distance information when the back appears.

  In the posture determination apparatus 1 according to the present embodiment, the determination images acquired by the determination image acquisition unit 10 are formed by the determination image classification unit 21 so as to be divided at intervals of 0.1 m. There is no need to form segmented images by distance. For example, the comforter is easily affected on the head side of the subject S, so that the interval between the segment images may be shortened, and on the other hand, the segment image may be formed by increasing the interval between the segment images. .

  Further, in the posture determination apparatus 1 of the present embodiment, the moving body detection unit 22 detects a moving body image for each segment image, and the combined moving body image forming unit 23 forms a combined moving body image while synthesizing from the head side of the subject S. However, although the posture determination is performed using the synthesized moving body image, it is not always necessary to perform the posture determination while combining the moving body images. For example, the moving body images of all the segment images may be synthesized first, and the posture determination may be performed while excluding the synthesized moving body of each segment from the synthesized video image of all the segments. Specifically, the synthesized moving body image that has been entirely synthesized is used, and the moving body image in the foot side segment is removed from the foot side of the subject S. In this case, since the comforter is included as a moving body image from the beginning, the degree of coincidence with the template image is low, but the degree of coincidence with the template image is high when the image of the comforter is removed. Can be determined as the posture of the subject S.

  Moreover, in the posture determination apparatus 1 of the present embodiment, a moving body image is detected from each segmented image by the background difference method or the posture determination of the target person S is performed by template matching. However, the present invention includes a depth value. Based on the determined image, a segmented image is formed for each fixed distance using the depth value, and posture determination is performed using the segmented image. Therefore, for example, a method other than the background subtraction method can be applied to detect a moving body image from each segmented image, and posture determination can also be performed by applying an image recognition method other than template matching for posture determination. Is also possible.

  In addition, in the posture determination apparatus 1 of the present embodiment, the input unit 2 constituting the determination image acquisition unit 10 is installed on the head side of the bed B in order to reduce the psychological burden on the subject S. It is also possible to obtain a determination image by installing it on the side or the ceiling of the room and perform posture determination by the posture determination means 20.

[Embodiment 2]
Next, the reporting system 100 according to the second embodiment will be described. The notification system 100 performs posture determination using the same function as the posture determination device 1 of the first embodiment, determines whether to perform notification based on the result of the posture determination, and based on the determination result. It is a system that reports.

  Specifically, the notification system 100 performs the posture determination of the subject using the determination image acquisition unit 110 formed by the depth value, and the divided image obtained by dividing the determination image into images of a certain distance using the depth value. Attitude determination means 120, notification determination means 130 for determining whether or not to report based on the determination result by the attitude determination means 120, notification means 140 for notification based on the determination result of the notification determination means, It has.

  Moreover, the hardware configuration of the notification system 100 is substantially the same as that of the posture determination apparatus 1 shown in FIG. 1, and different points will be described as appropriate, and the same reference numerals will be used for the same configuration and detailed description. Is omitted.

  The determination image acquisition unit 110 and the posture determination unit 120 are the same as the determination image acquisition unit 10 and the posture determination unit 20 of the posture determination apparatus 1 according to the first embodiment.

  The notification determination unit 130 is configured by the control unit 3 and the storage unit 4, and determines whether to perform notification based on the determination result by the posture determination unit 120. Specifically, for example, the notification determination unit 130 determines that the notification is performed when it is determined that the subject S is in the posture P3, and determines that the notification is not performed in the posture P1 or the posture P2.

  Note that whether or not to make a report is determined based on a determination condition set in advance. Further, since the degree of risk depending on the posture is different for each target person S, this determination condition may be different for each target person S. Such a determination condition can be set by, for example, an input device (such as a keyboard or a touch panel) that constitutes the input unit 2 other than the camera 2a and the depth sensor 2b included in the reporting system 100.

  The reporting unit 140 is mainly configured by the output unit 5, and reports based on the determination result of the reporting determination unit 130. Specifically, the reporting unit 140 performs reporting by display on a monitor or sound by a speaker when it is determined by the reporting determination unit 130 that reporting is to be performed.

  The notification system 100 having such a configuration will be described based on an actual usage example. FIG. 14 is a diagram illustrating an installation example of the notification system 100. The use example shown in FIG. 14 assumes use in a facility that needs to watch over the target person S on a bed, such as a hospital or a care facility for elderly people.

  This facility has a plurality of rooms R (R1 to R3), and a bed B is installed in each room. In FIG. 14, only one bed B is shown in each room R, but a plurality of beds may be installed. In FIG. 14, the number of rooms is only three, but the reporting system 100 can be used even with a larger number of rooms.

  In the bed B of each room R, the input unit 2 constituting the determination image acquisition unit 110 is installed. Further, the posture determination means 120, the notification determination means 130, and the notification means 140 constituting the notification system 100 are installed at a station D where a nurse or a caregiver waits. Note that the reporting unit 140 may be installed not only in the station D but also in each room R.

  For example, when the posture P2 that is the determination condition of the notification for the subject S in the room R2 is determined, the notification means 140 of the room R2 notifies the person R, and the person R in the room R2 or near the room R2. It is possible to inform a person that the subject S is in a dangerous posture P2.

  Here, the terminal 200a installed in the station D shown in FIG. 14 is a terminal connected to the reporting system 100 via the communication unit 6 constituting the reporting system 100.

  Further, the terminal 200b is a terminal installed in a station D ′ different from the station D, for example, on a different floor from the station D or in a place away from the station D. The terminal 200b is also a terminal connected to the notification system 100 via the communication unit 6 constituting the notification system 100.

  The portable terminal 200c is a terminal that is carried by, for example, a nurse or a caregiver working in a facility. The terminal 200c is also a terminal connected to the notification system 100 via the communication unit 6 constituting the notification system 100.

  The reporting system 100 can use the display unit and the voice output unit of the terminals 200 a, 200 b, and 200 c connected via the communication unit 6 as the reporting unit 140. And the reporting system 100 can report the determination result by the report determination means 130 to a plurality of nurses and caregivers by using the terminals 200a, 200b, and 200c as the notification means 140.

  In addition, such a reporting system 100 reports, for example, the subject S in the room R2 to a specific nurse and reports the subject S in the room R3 to another nurse. It is also possible.

  As the terminals 200a, 200b, and 200c, not only terminals such as notebook computers, tablets, and smartphones, but also dedicated terminals made exclusively for reporting can be used.

  As described above, the notification system 100 according to the present embodiment acquires the determination image formed by the depth value as in the posture determination device 1 according to the first embodiment, and divides the image into images for each predetermined distance using the depth value. Since the posture determination is performed based on the segmented image, the posture determination can be performed without the influence of the comforter used by the subject on the bed. In addition, since the notification system 100 determines whether or not to report based on the determination result of the posture determination and performs notification based on the determination result, many determination target persons S such as hospitals and care facilities. Very suitable for use in facilities with

  Note that the posture determination apparatus of the first embodiment and the notification system 100 of the second embodiment mainly prevent the fall of the subject S from the bed B by mainly judging the posture of the subject S on the bed. Although it is used, it can be used for purposes other than preventing falling from the bed B. For example, if the same posture is continued on the bed for a long time, there is a risk of pressure ulcers called bedsores. Therefore, the posture determination apparatus and the notification system of the present invention notify the nurse or caregiver that it is necessary to change the posture of the subject S when it is determined that the same posture continues for a long time. Can also be used.

  In addition, the posture determination apparatus according to the first embodiment and the notification system 100 according to the second embodiment determine the posture of the target person S. However, the target person S is not limited to a human being, but a person. Other animals and objects are also widely included and can be rephrased as objects. And the word posture determination is not limited to the determination of the posture on the bed, but also includes determinations other than on the bed and other than the posture, and can be rephrased as shape determination. It is very flexible.

DESCRIPTION OF SYMBOLS 1, 1A ... Attitude determination apparatus 2 ... Input part 6 ... Communication part 10, 10A, 110 ... Determination image acquisition means 20, 20A, 120 ... Attitude determination means 21 ... Determination image classification means 22, 21A ... Moving object detection means 23 ... Composition Moving body image forming means 24, 22A ... Template matching means 30, 30A ... Determination result output means 100 ... Notification system 130 ... Notification determination means 140 ... Notification means 200a, 200b, 200c ... Terminal S ... Subject B ... Bed P1 ... Posture 1
P2 ... Attitude 2
P3 ... Attitude 3
D, D '... Station R ... Room

Claims (4)

Determination image acquisition means for acquiring a determination image formed by a depth value;
Posture determination means for performing posture determination of a subject by a divided image obtained by dividing the determination image into images at fixed distances using the depth value;
An attitude determination device comprising: The posture determination means includes
A moving object detecting means for detecting a moving object image having movement from the segmented image;
A combined moving body image forming means for combining the moving body images of the different segment images to form a combined moving body image;
With
The posture determination apparatus according to claim 1, wherein the posture of the subject is determined from the synthesized moving body image. The posture determination means includes
The posture determination apparatus according to claim 2, further comprising a template matching unit that determines a posture of the synthesized moving body image by performing template matching using a template image. Determination image acquisition means for acquiring a determination image formed by a depth value;
Posture determination means for performing posture determination of a subject by a divided image obtained by dividing the determination image into images at fixed distances using the depth value;
Notification determination means for determining whether to perform notification based on the determination result by the posture determination means;
A reporting means for reporting based on a determination result of the reporting determination means;
A reporting system comprising: