JP2019195395A - Risk degree estimation system - Google Patents

Abstract

To provide a risk degree estimation system capable of acquiring risk of accident due to fall or inversion of a patient on a bed on a medical site, in a contactless and unconstrained manner.SOLUTION: A risk degree estimation system 1 comprises: a right above image acquisition part 10 acquiring a right above image of an object person S on a bed which is a risk degree estimation object, the right above image being formed on the basis of a depth value image formed on the basis of the depth value; a risk degree estimation part 20 estimating a risk degree on the basis of the right above image; and an estimation result output part 30 outputting the estimation result.SELECTED DRAWING: Figure 3

Description

  The present invention relates to a risk estimation system for acquiring a risk of an accident due to a patient falling or falling on a bed in a medical field in a non-contact / non-restraint manner.

  Various problems relating to such accidents are known because accidents in which hospitalized patients or elderly people fall from the bed and fall are a problem. As an example, there is known a device that arranges a sensor for detecting a load next to a bed and reports that a patient or the like has fallen when the sensor detects the load.

  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. .

  Also, there is a known apparatus that acquires an image of a patient on a bed and estimates the posture based on the image as disclosed in Patent Document 2, instead of using a load sensor or attaching a sensor to the patient. It has been.

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

  However, an apparatus that arranges a load sensor beside the bed has many problems such as many false alarms and notification after the load is detected, so that it is difficult to prevent the fall.

  Moreover, the method of attaching a sensor to the patient himself / herself in Patent Document 1 makes the sensor attached to the patient feel uncomfortable, and the sensor can be removed without permission, or attached to an appropriate position due to the influence of surgery or the like. There are some difficulties.

  Moreover, since the apparatus which acquires a patient's image and performs posture determination like patent document 2 is non-contact and non-restraint, it gives a patient discomfort during use, or cancels itself. Although there are merits such as not being able to be performed, there are problems with the accuracy of judgment, etc., and it is still not sufficient for use in actual sites. In particular, in Patent Document 2, an image of a patient is placed beside the patient's head and acquired. For example, when it is desired to estimate the degree of danger including the situation around the bed, it is difficult to estimate the image from the lateral direction sufficiently because of the problem of the viewing angle.

  SUMMARY OF THE INVENTION An object of the present invention is to provide a risk estimation system that can acquire the risk of an accident due to a patient falling or falling on a bed in a medical field without contact and without restraint.

  In order to solve the above-described problem, the risk estimation system of the present invention is formed on the basis of a depth value image formed by depth values. A risk level estimation unit that estimates a risk level based on an image directly above and an estimation result output unit that outputs a result of estimation by the risk level estimation unit are provided.

Further, the risk estimation system of the present invention is divided into determination layer dividing means for dividing the risk estimation unit into determination layers for performing risk estimation using the depth value from the immediately above image. And a target layer determining means for determining a target layer from among the determination layers, wherein the risk level is estimated based on the target layer determined by the target layer determining means.
In the risk estimation system according to the present invention, the determination layer dividing unit includes a cube dividing unit that divides a plurality of cube regions.

In the risk estimation system of the present invention, the risk estimation unit further calculates a risk based on a position specifying unit for specifying the position of the object and a position specified by the position specifying unit. And a risk level calculating means.
In the risk estimation stem of the present invention, the directly above image acquisition unit includes an directly above image converting unit that converts the depth value image into the directly above image.

  The risk level estimation system of the present invention is configured to acquire a directly above image of a risk level estimation object formed based on a depth value image formed based on a depth value, and to estimate the risk level based on the directly above image. Therefore, it is possible to acquire the risk of an accident due to a fall or fall of a patient on a bed in a medical field without contact and without restraint. There has never been a risk level estimation system that estimates the level of risk using an immediately above image formed based on a depth value image. Further, by using the image directly above, it is possible to easily estimate the degree of risk including the situation around the bed.

  In the risk estimation system of the present invention, the risk estimation unit is configured to divide the determination layer into determination layers for performing risk estimation, and to determine a target layer from the divided determination layers. And a layer determining means, and the risk is estimated based on the layer of interest. By such a method, it is possible to estimate the degree of risk using the image directly above.

  In addition, since the determination layer dividing unit includes a cube dividing unit that divides into a plurality of cube regions, the region is divided not only in the depth direction (Z-axis direction) but also in the plane region (X-axis direction, Y-axis direction). Thus, the risk level can be estimated with higher accuracy.

  In addition, since the risk level estimation unit further includes a position specifying unit for specifying the position of the object and a risk level calculating unit for calculating the risk level based on the specified position, the risk level estimation unit is more accurate. It is possible to estimate the risk well.

  In addition, since the overhead image acquisition unit includes a overhead image conversion means for converting the depth value image into the overhead image, the overhead image can be acquired even from the depth value image obtained by photographing the risk estimation target object from an oblique direction. Can do. Note that when photographing from an oblique direction, an input unit for acquiring a depth value image can be installed within a range that can be converted into an image directly above, so that there are a great number of options for the installation position of the input unit.

It is a block diagram which shows the hardware constitutions of the risk estimation system of Embodiment 1. FIG. It is a block diagram which shows the function structure of the risk estimation system of Embodiment 1. It is the figure which showed the usage example of the danger level estimation system of Embodiment 1, and the concept of risk level estimation. It is a block diagram which shows the function structure of the risk estimation system of Embodiment 2. FIG. (A) is the figure which showed the usage example of the danger level estimation system of Embodiment 2, and the concept of risk level estimation, (B) is the figure which showed the symbol used in risk level estimation according to the usage example conceptually. It is. It is an image figure of the coordinate conversion which converts the depth value image of a diagonal direction into an image directly above. It is the figure which showed the specific example which extracted the division with the maximum data number for every layer of Z-axis in risk estimation. It is the figure which showed the specific example which added together the data of the middle layer ML and lower layer LL when paying attention to the middle layer ML and the lower layer LL as a focused layer in risk estimation. It is the figure which showed the example of a setting of the weight of risk in verification of risk estimation. It is the figure which showed the time series data in verification of risk estimation. It is the figure which arranged the depth value image in verification of risk estimation in order of high risk.

  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, the basic configuration of the risk level estimation system 1 of the present embodiment will be described with reference to FIG. 1, FIG. 2, and FIG. FIG. 1 is a block diagram showing a hardware configuration of the risk level estimation system 1. FIG. 2 is a block diagram showing a functional configuration of the risk level estimation system 1. FIG. 3 is a diagram illustrating a usage example of the risk level estimation system 1 and a concept of risk level estimation.
As shown in FIG. 1, the risk estimation stem 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 a depth value image formed by depth value information for a range including the target person S who performs risk level estimation using the risk level estimation system 1. It is what is called. In the present embodiment, specifically, the input unit 2 has a camera 2a for acquiring an image including the target person S on the bed B, and depth information regarding the depth to the subject in the image captured by the camera 2a. And a depth sensor 2b for acquiring 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. If sound information can also be used for risk estimation, 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 provides information to the risk estimation system 1 such as inputting the name of a person who is a target of risk estimation, for example. An input device such as a keyboard or a touch panel for inputting may be included.

  The control unit 3 includes a CPU, a RAM, a ROM, and the like, and executes various processes associated with control of each unit and risk estimation based on a stored program. The storage unit 4 is a storage device such as a DRAM or HDD, and stores various data such as temporarily storing image data acquired by the input unit 2. The output unit 5 includes a display device, a speaker, and the like, and outputs an estimation result or the like based on risk estimation using video or 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 risk level estimation system 1 can output the risk level estimation results to other devices. The risk estimation system 1 may be configured to be connectable to an existing nurse call system by the communication unit 6.
As shown in FIG. 2, the risk level estimation system 1 includes a top image acquisition unit 10, a risk level estimation unit 20, and an estimation result output unit 30.

  The overhead image acquisition unit 10 is mainly configured by the input unit 2, and uses the input unit 2 to obtain a depth value based on a depth value obtained by observing the overhead image observed from above so as to overlook the target person S who performs risk estimation. Acquire based on images. In order to acquire such a directly above image, the directly above image acquiring unit 10 in the present embodiment is configured to include only the depth value image acquiring unit 11 including the input unit 2. Accordingly, the upper image acquisition unit 10 installs the input unit 2 as the depth value image acquisition unit 11 above the subject, and acquires the depth value image itself obtained from the upper input unit 2 as the upper image. .

  The risk level estimation unit 20 is mainly configured by the control unit 3 and the storage unit 4, and uses the depth value included in the upper image based on the upper image acquired via the upper image acquisition unit 10. Estimate the risk level. The risk estimation unit 20 according to the present embodiment is characterized in that the immediately above 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 risk level estimation, the risk level estimation unit 20 includes a determination layer dividing unit 21 and a target layer determination unit 22.

  Although details will be described later together with a specific example, the determination layer dividing unit 21 divides the determination layer for determination of risk using the depth value from the immediately above image. The target layer determination unit 22 determines a target layer from among the divided determination layers. Then, the risk level estimation unit 20 estimates the risk level based on the target layer determined by the target layer determination unit 22.

  That is, the risk level estimation unit 20 according to the present embodiment divides the overhead image into images at a certain distance along the depth direction (Z-axis direction) based on the information on the depth value of the overhead image. Each image is formed as a determination layer. Then, the risk estimation unit 20 determines a determination layer in which the appearance of the subject S is seen as a target layer depending on whether or not a part of the subject S appears in each determination layer, and the focus Based on the stratum, it is estimated whether or not the subject person S is in danger of falling.

  The estimation result output unit 30 is mainly configured by the output unit 5 and outputs the risk level of the subject S estimated by the risk level estimation unit 20 using a monitor or the like. As for the output of the estimation result, for example, when the risk of falling is high, it is displayed in characters or output by voice so that the risk is high.

  Next, the risk estimation system 1 having such a configuration will be described based on an actual usage example. FIG. 3 is a diagram showing a concept of risk estimation together with a usage example of the risk estimation system 1. The risk level estimation system 1 estimates the risk level of the subject S using the bed B on the bed B using the directly above image acquired by the directly above image acquiring unit 10.

  As described above, KINECT (registered trademark) manufactured by Microsoft Corporation was used as the distance image sensor of the input unit 2 constituting the image acquisition unit 10 directly above. This KINECT has a function of acquiring a depth distance in a 512 × 424 [pixel] plane. The input unit 2 needs to photograph the subject S from above in order to obtain an image directly above, and is installed on the ceiling C of the room where the bed B is placed.

  Further, in the present embodiment, the hardware configuring the risk estimation unit 20 is also arranged in the room R where the bed B is installed, and the hardware configuring the estimation result output unit 30 is the room It is arranged in a room different from R (for example, a nurse station).

  In addition, the danger level estimation system 1 can also arrange | position the hardware which comprises the estimation result output part 30 in the room R in which the bed B is installed. In this case, since the estimation result output unit 30 needs to notify that the subject S is in a dangerous state, the estimation result output unit 30 is preferably output by sound such as an alarm sound. Moreover, by installing in the room R, the subject S himself / herself can be made aware that it is dangerous.

  In addition, the risk level estimation system 1 arranges the hardware configuring the overhead image acquisition unit 10 in the room R, and sets the hardware levels of the risk level estimation unit 20 and the estimation result output unit 30 in a room different from the room R. It is also possible to adopt a configuration of arranging.

  The data of the directly above image acquired by the directly above image acquiring unit 10 is sent to the risk level estimating unit 20 of the risk level estimating system 1. Then, the risk level estimation unit 20 estimates the risk level of the target person S based on the directly above image acquired by the top image acquisition unit 10. Here, a specific example of risk estimation in the risk estimation unit 20 in the present embodiment will be described.

  First, the directly above image acquired by the directly above image acquiring unit 10 is formed by a depth value image based on a depth value. For this reason, depth information is included in the image directly above. Therefore, the determination layer dividing unit 21 uses this depth value to divide into determination layers. Specifically, the image immediately above is divided into images at a certain distance in the depth direction (Z-axis direction), and each image at each certain distance is formed as a determination layer.

  FIG. 3 shows an example in which the determination layer is divided into two layers, a lower layer LL and an upper layer UL. The lower layer LL is a determination layer composed of an image of 0.1 to 0.9 m with the bed B surface as a reference (0 m). The upper layer UL is a determination layer made of an image of 0.9 to 1.4 m.

  In addition, about the range of lower layer LL, it was set from 0.1 m instead of the standard 0 m of the bed B surface. The reason for this is that if it is from the surface of the bed B (0 m), it will be affected by the comforter, wrinkles of the sheets, sinking of the mattress, and the like. And since the influence of the thickness of the comforter was particularly large and the thickness of the comforter was approximately 0.1 m or less, the lower layer LL was set to 0.1 m. Therefore, if the comforter to be used is thicker, for example, it is possible to start from 0.2 m. However, if the value is too large, it will be impossible to detect the posture of the supine position described later. Therefore, in consideration of the thickness of the human body that is the subject S, it is preferable to set the value from 0.1 m.

  Then, the target layer determination unit 22 determines the target layer from the determination layers divided into the lower layer LL and the upper layer UL. The target layer can be determined as follows. First, it is determined by whether or not the number of data for forming an image directly above the upper layer UL is greater than or equal to a predetermined value (threshold value). If the upper layer UL has a number of data equal to or greater than the threshold, the upper layer UL is determined as the target layer.

  If the upper layer UL does not have the number of data equal to or greater than the threshold value, the lower layer LL is focused next, and the determination is made based on whether or not the number of data for forming the image directly above the lower layer LL is equal to or greater than the threshold value. If the lower layer LL has a data number equal to or greater than the threshold, the lower layer LL is determined as the target layer.

  This point will be described with reference to FIG. FIG. 3 shows three typical postures assumed by the subject S in the bed B. These three postures are the posture indicated as P1: the supine position on the bed B, the posture indicated as P2, the sitting position on the bed B, and the posture indicated as P3: standing from the bed B. Note that the height from the floor of the room R to the surface of the bed B is 0.47 m.

  For example, if the posture of the subject S is the posture P1 or P2, there is no image data of the subject S in the image directly above the upper layer UL, and therefore the upper layer UL does not have the number of data equal to or greater than the threshold value. On the other hand, if the posture of the subject S is the posture P3, the image of the head and shoulders of the subject S exists in the image directly above the upper layer UL. Accordingly, the upper layer UL has a number of data equal to or greater than the threshold value, and the target layer determination unit 22 determines the upper layer UL as the target layer.

  Further, if the posture of the subject person S is the posture P1 or P2, there is no image data of the subject person S in the image directly above the upper layer UL, but since the image of the subject person S exists in the lower layer LL, the threshold value or more is exceeded. There will be a number of data. Accordingly, the target layer determination unit 22 determines the lower layer LL as the target layer.

Thus, the risk level estimation unit 20 determines the target layer and estimates the risk level. For example, if the target layer is the upper layer UL, it can be considered that the subject S stands up from the bed B and is in a standing posture. Therefore, it can be said that the risk of falling is very high. For this reason, the risk level estimation unit 20 can estimate that the risk level of falling is large. Further, if the target layer is the lower layer LL, it can be said that the risk of falling is low because the subject S is in a supine or sitting position on the bed B. Therefore, the risk level estimation unit 20 can estimate that the risk level of falling is small.
Then, the estimation result output unit 30 outputs the result of the risk level estimation of the subject S estimated by the risk level estimation unit 20 by the estimation result output unit 30 via a monitor or the like.

  As described above, the risk level estimation system 1 acquires an image directly above the target person S on the bed, which is a risk level estimation object, formed based on the depth value image formed by the depth value. The configuration includes a unit 10, a risk level estimation unit 20 that estimates the level of risk based on the image directly above, and an estimation result output unit 30 that outputs the estimation result.

  For this reason, it is possible to perform risk estimation with higher accuracy than in the case of using a depth value image photographed from the lateral direction (Y-axis direction) of the bed B as in the conventionally known Patent Document 2. Become. That is, in the case of a depth value image from the horizontal direction, a depth value image cannot be obtained if it is on the extension line of the subject S. For example, when the nurse is in the vicinity of the bed B, the risk is usually very low, or it is not necessary to estimate the risk. However, in the image from the horizontal direction, it is difficult to estimate the degree of risk because the nurse who is on the extension line of the subject S cannot be recognized. Also, from the lateral direction, when the information around the bed B is known, it cannot be completely captured due to the problem of viewing angle. Moreover, when there are objects (for example, a shelf and a refrigerator) placed side by side, it is difficult to obtain both information and information about the height thereof. As described above, in the risk level estimation using only the depth value image from the horizontal direction, information to be obtained is limited. Therefore, it is difficult to estimate the risk level with high accuracy.

  Further, the risk level estimation system 1 includes a determination layer dividing unit 21 that divides the risk level estimation unit 20 into determination layers including a lower layer LL and an upper layer UL for risk level estimation using a depth value from an immediately above image. The target layer determining means 22 for determining the target layer from the judgment layers, and the risk level is estimated based on the target layer determined by the target layer determining means 22. For this reason, the danger level estimation system 1 can acquire the risk of the accident of the subject S in the bed B without contact and without restraint.

  The risk level estimation system 1 according to the present embodiment has a configuration in which the risk level estimation unit 20 includes the determination layer division unit 21 and the target layer determination unit 22, but when such a configuration is not provided. For example, it is conceivable to specify the shape at a predetermined distance or the position on the bed B from the image directly above, specify the posture of the subject S by pattern matching or the like from the information, and perform the risk estimation.

  In addition, in the target layer determination unit 22, when the number of data for forming an image directly above the lower layer LL is not greater than or equal to the threshold value, the risk level is estimated as the subject S is absent (for example, there is no risk of falling because of the absence) Can be estimated). In addition, the target layer determination means 22 determines whether the number of data for forming an image directly above the determination layer is greater than or equal to a threshold value, such as a bed B, a fence installed around the bed B, and the like. This is because various objects other than the person S exist to prevent an influence when an object other than the object person S is included in the image directly above. For example, when KINECT (registered trademark) that acquires 512 × 424 [pixel] data is used in the input unit 2, it may be about 1500 as an example of the threshold value of the number of data.

[Embodiment 2]
Next, a risk estimation system 1A that is another embodiment will be described. FIG. 4 is a block diagram showing a functional configuration of the risk estimation stem 1A. FIG. 5 is a diagram showing a usage example of the risk level estimation system 1A and the concept of risk level estimation.

  The risk level estimation system 1A of the present embodiment has the same hardware configuration as the risk level estimation system 1 of the first embodiment shown in FIG. 1, and includes an input unit 2, a control unit 3, a storage unit 4, and an output. Unit 5 and communication unit 6. As shown in FIG. 4, the risk level estimation system 1 </ b> A includes a top image acquisition unit 10 </ b> A, a risk level estimation unit 20 </ b> A, and an estimation result output unit 30 </ b> A.

  The overhead image acquisition unit 10 </ b> A is mainly configured by the input unit 2. The overhead image obtained by observing the target person S who performs risk estimation from above is displayed as a depth value image based on the depth value using the input unit 2. Get based on.

  Here, in the risk level estimation system 1 of the first embodiment, the overhead image acquisition unit 10 includes only the depth value image acquisition unit 11 and is installed on the ceiling C of the room R above the subject S. The depth value image itself obtained from the above was acquired as a directly above image.

  By the way, in order to acquire an image directly above, the ideal location for the input unit 2 is the ceiling C of the room R, particularly the center of the bed B, as in the first embodiment. However, installing the input unit 2 on the ceiling C of the room R may cause the input unit 2 to fall onto the subject S, and may cause the subject S to feel uncomfortable when the input unit 2 enters the field of view. It is not so preferable because there is.

  Therefore, in the present embodiment, as shown in FIG. 5, the input unit 2 is installed on the head side of the bed B that is difficult to enter the field of view of the subject S. For this reason, in this embodiment risk level estimation system 1A, as a configuration of the overhead image acquisition unit 10A, a depth value image acquisition means 11A and an overhead image conversion means 12A are provided.

  The depth value image acquisition unit 11A acquires a depth value image based on the depth value, similarly to the depth value image acquisition unit 11 of the first embodiment. At this time, in this embodiment, the depth value image is an image obtained by photographing the bed B from an oblique direction. Therefore, in order to obtain an image directly above the risk level, it is necessary to convert the image into an image taken from directly above the subject S using the depth value image taken from the oblique direction.

  Therefore, the upper image conversion unit 12A converts the depth value image acquired by the depth value image acquisition unit 11A into an upper image. A specific method for converting the depth value image into the image directly above will be described later together with a specific example.

  The risk level estimation unit 20A is mainly configured by the control unit 3 and the storage unit 4, and based on the directly above image acquired via the directly above image acquiring unit 10A, the risk level estimation unit 20A uses the depth value included in the directly above image. Estimate the risk of S. As an example for realizing such risk level estimation, the present embodiment includes a cube division unit 21A, a target layer determination unit 22A, a position specifying unit 23A, and a risk level calculation unit 24A.

  The cube dividing means 21A uses the depth value from the immediately above image to divide into grid-like areas for estimating the degree of risk. Here, the determination layer dividing unit 21 according to the first embodiment only divides in the depth direction (Z-axis direction) like the lower layer LL and the upper layer UL. On the other hand, the cube dividing means 21A of this embodiment not only divides in the Z-axis direction, but also divides in the horizontal direction (X-axis direction) and vertical direction (Y-axis direction) of the bed B as shown in FIG. Thus, the configuration is such that a plurality of cubic regions are divided.

In this way, by dividing not only in the Z-axis direction but also in the X-axis direction and the Y-axis direction, it is possible to estimate the degree of danger including where the subject S is on the bed B. Therefore, the risk level estimation system 1A can perform risk level estimation with higher accuracy.
The target layer determination means 22A determines the target layer in the Z-axis direction for estimating the risk level from the divided cubic region.

  The position specifying unit 23A specifies the position where the target person S exists. In the present embodiment, based on the target layer determined by the target layer determination unit 22A, the position of the target person S is specified by estimating the section of the cubic region where the target person S exists using the position of the center of gravity. .

  The risk level calculation means 24A calculates the risk level of the subject S. In the present embodiment, the degree of risk is calculated using the position of the center of gravity by the position specifying means 23A. By calculating such a risk level, it is possible to perform a risk level estimation with higher accuracy than the risk level estimation unit 20 of the first embodiment.

Further, by using the position of the center of gravity by the position specifying means 23A, it is possible to estimate the degree of risk reflecting the difference in the degree of danger depending on the position of the bed B. That is, even on the bed B, the risk of falling differs depending on the position on the bed B, such as the possibility of the bed B falling over and the possibility of falling down at the bedside. Therefore, it is possible to calculate the degree of risk based on the position and weight of the target person S by weighting the risk to each section.
The estimation result output unit 30A is mainly configured by the output unit 5, and outputs the risk level of the subject estimated by the risk level estimation unit 20A using a monitor or the like.

  As described above, in the risk level estimation system 1A of the present embodiment, the upper image acquisition unit 10A first acquires the depth value image obtained by photographing the bed B from the oblique direction by the depth value image acquisition unit 11A, and then the upper image. The image is converted into the image immediately above by the conversion means 12A. Then, the risk estimation system 1A is divided into a plurality of sections in the X, Y, and Z-axis directions based on the depth value information by the cube dividing means 21A by the risk estimating section 20A, and the target layer determining means 22A performs Z The target layer in the axial direction is determined, the position of the target person S is specified by the position specifying means 23A, the risk is calculated by the risk calculating means 24A, the risk is estimated, and the output of the estimation result is estimated. This is performed by the output unit 30A.

Next, the risk estimation system 1A having such a configuration will be described based on an actual use example of FIG. In FIG. 5B, symbols used below are conceptually shown according to usage examples. The upper layer UL is Z = 2, the middle layer ML is Z = 1, the lower layer LL is Z = 0, and FIG. 5B shows an example of the lower layer LL (Z = 0). Further, d is the boundary line of each pixel, n is the number of data in the section, and w is the weight of the fall risk.
The risk level estimation system 1A estimates the risk level of the subject S using the bed B on the bed B using the directly above image acquired by the directly above image acquiring unit 10A.

As the distance image sensor of the input unit 2 constituting the immediately above image acquisition unit 10A, as in the first embodiment, KINET (registered trademark) manufactured by Microsoft Corporation was used. Moreover, this input part 2 is installed in the bed head side so that it may become difficult to enter into the view field of the subject S.
The conditions in this embodiment are as follows.
Bed B size (W x D x H) 0.9 x 2.1 x 0.5 [m]
Inclination angle θ 32.4 [°] of the input unit 2
Height from the ground to the input unit 2 1.95 [m]
Height from bed B to input 2 1.45 [m]

  Since the depth value image acquired by the depth value image acquiring unit 11A of the directly above image acquiring unit 10A is data captured from an oblique direction, it is necessary to convert the depth value image into an image directly above the center of the bed B. . Therefore, in the present embodiment, the depth value image obtained from the oblique direction is converted into the directly above image by the directly above image converting means 12A. For conversion to an image directly above, the acquired data is projected using equation (1).

  Here, θ is the inclination of the input unit 2 with respect to the ground, and is 32.4 ° in this embodiment. The conversion of Expression (1) is performed on all the 512 × 424 [pixel] distance information constituting the depth value image. A top image is obtained from this conversion result.

  In addition, it supplements about the conversion to a direct image. FIG. 6 shows an image diagram of coordinate conversion by equation (1). In order to convert the depth information obtained from the input unit 2 into information photographed from directly above, the ZY coordinate system may be converted to the Z′Y ′ coordinate system. Since the Z′Y ′ coordinate is inclined by an angle 90−θ [°] from the ZY coordinate, if the coordinate data P is multiplied by a rotation matrix, it can be converted to Z′Y ′ coordinate axis data P ′. From this, it is possible to convert the image directly above according to the equation (1).

The data of the directly above image by the directly above image acquiring unit 10A is sent to the risk degree estimating unit 20A. First, the cube dividing unit 21A divides the cube into a plurality of cubic regions in the X, Y, and Z axis directions based on the depth value information. By the way, the data of the image immediately above has a lot of distance information of 512 × 424 pieces as in the depth value image before conversion. Therefore, if all the distance information is used as it is, it is difficult to estimate the risk level in real time. For this reason, dividing the data of the immediately above image into cubic area sections also leads to a reduction in dimension of the data and also facilitates risk level estimation in real time.
A specific method for dividing the data into cube sections in the cube dividing means 21A for reducing the data dimension will be described below.
[1] The range of the bed B is selected in the data of the immediately above image. In addition, when the state of the person around the bed B is to be captured, a range outside the bed B is also set.

[2] Data is divided into a grid within the selected range. The division is performed for all of the X axis, the Y axis, and the Z axis. As shown in FIG. 5, the reason why the Z-axis is made up of the lower layer LL, the middle layer ML, and the upper layer UL is to enable the determination of the supine position P1, the sitting position P2, and the standing position P3. The sitting position P2 is the posture before the transition to the standing position P3, and in order to prevent the subject S from falling, it is better to be able to determine the posture of the sitting position P2 in addition to the standing position P1 and the standing position P3. is there.

[3] Divide each section into a plurality of sections in the cubic region, with threshold _values for dividing each section being d ^X _x , d ^Y _y , and d ^Z _z . In this embodiment, the image is divided into 3 × 3 × 3 27 sections.
[4] In the present embodiment,

Let n (x, y, z) be the number of data existing in the range. Then, the number of data n (x, y, z) of all sections is obtained.

[5] A partition having the maximum number of data is extracted for each layer of the Z axis (upper layer UL, middle layer ML, lower layer LL). FIG. 7 shows a specific example in which a section having the maximum number of data is extracted for each layer of the Z axis. A section having the maximum number of data for each layer is indicated by a gray section in each layer.
And
Maximum number of upper layer UL data is n _max (2)
The maximum number of data in the middle layer ML is n _max (1)
The maximum number of data in the lower layer LL is n _max (0)
And

  Next, the target layer is determined by the target layer determination means 22A. Here, in the present embodiment, the section of the cubic region where the target person S exists is specified by the target layer determination unit 22A and the next position specifying unit 23A. Below, the method to determine the division where the subject S exists is demonstrated.

[1] First, in order to determine which layer the subject S is in, it is compared whether the maximum number of data in each layer exceeds a predefined threshold. At this time, the lower layer LL has a threshold value n _th (0), the middle layer ML has a threshold value n _th (1), and the upper layer UL has a threshold value n _th (2). In this embodiment, n _th (0) = n _th (1) = n _th (2) = 1500 is set. At this time, in the specific example shown in FIG. 7, the middle layer ML and the lower layer LL exceed the threshold.
[2] From this result, the target layer is determined according to the following.
[A] IF n _max (2) ≧ n _th (2) THEN Focus on the upper layer UL
To ELSE b) [b] IF n _max (1) ≥ n _th (1) THEN Pay attention to middle layer ML and lower layer LL
To ELSE c) [c] IF n _max (0) ≥ n _th (0) THEN Focus on middle layer ML and lower layer LL
Absence of ELSE (not seeing any layer)

  Here, in the specific example of FIG. 7, [b] is selected. In addition, regarding the middle layer ML and the lower layer LL, the middle layer ML and the lower layer LL may not be focused together as described above, but may be focused separately. However, if attention is paid separately to the middle layer ML and the lower layer LL, the data becomes discontinuous when the layer of interest is switched from the middle layer ML to the lower layer LL or the lower layer LL to the middle layer ML. It is preferable to pay attention. However, if the threshold value is determined after the middle layer ML and the lower layer LL are combined, there is a possibility that correct determination cannot be made due to noise, so the determination of the middle layer ML and the lower layer LL is performed as described above. It is preferable to divide into two.

[3] When focusing on the middle layer ML and the lower layer LL as the target layer in [2], as shown in FIG. 8, when adding the data of the middle layer ML and the lower layer LL and focusing on the upper layer UL, n (x, y, 2) is used as it is. When this is expressed by an equation, the equation (3) is obtained.

[4] In the layer of interest, the x, y, and z of the section where n ′ (x, y, z) is the maximum are set to x _max , y _max , and z _max , respectively.
[5] Based on the maximum number of sections, the center-of-gravity coordinates g ^{X on} the X-axis, the center-of-gravity coordinates g ^{Y on} the Y-axis, and the center-of-gravity coordinates g ^Z on the Z-axis are obtained by Expression (4).

The obtained barycentric coordinates are defined as the position of the subject S.

Next, the risk level calculation means 24A calculates the risk level of the target person S based on the identified position of the target person S. Specifically, using the calculated center-of-gravity coordinates g ^X , g ^Y , and g ^Z , the following procedure is used.
[1] First, the weight of the fall risk is set for each section of the cubic region. _Let w _x , w _y , and w _z be the weights of the fall risk for each axis.
[2] From the barycentric coordinates g ^X and the following equation (5), the weight of the degree of risk to be used (hereinafter referred to as a fall risk as appropriate) is determined.

[3] Then, the fall risk r ^X in the X-axis direction is obtained by the equation (6) or the equation (7). In addition, the fall risks r ^Y and r ^{Z of the} Y axis and Z axis are obtained in the same manner.

[4] The total fall risk r (r = r ^X + r ^Y + r ^Z ) of the X axis, the Y axis, and the Z axis is the final fall risk in this embodiment.

Here, the result of verifying the usefulness of the risk estimation by the above method in the risk estimation unit 20A will be described. In order to verify the usefulness, the present inventor performed verification by an actual patient in an actual hospital. The test subjects were adult males in hospital. Experimental conditions are
Bed B size (W x D x H) 0.9 x 2.1 x 0.5 [m]
Inclination angle θ 32.4 [°] of the input unit 2
Height from the ground to the input unit 2 1.95 [m]
Height from bed B to input 2 1.45 [m]
Sampling interval 1.0 [s]
Threshold 1500 for determining the layer of interest (all upper layer UL to lower layer LL)
It is. Moreover, the weight of each division is shown in FIG.

  This time, in addition to the bed B, the X-axis measurement range of the upper layer UL was extended to the outside of the bed B in order to capture the state of the subject S outside the bed B (standing posture, etc.). The evaluation was performed by comparing the risk value calculated by the risk estimation system 1A with the weight of the fall risk outside the bed being 3, and the posture of the subject S at that time.

  And the time series data of the fall risk derived | led-out from experimental conditions are shown in FIG. In addition, FIG. 11 shows a result of arranging the depth value images (images showing the state of the subject S) at the points (a) to (l) in descending order of the degree of risk. In each of the images (a) to (l) in FIG. 11, the upper left numerical value is the calculated degree of risk, and the upper right numerical value is the time (s) at each point in FIG. 10.

  Looking at this result, the fall risk when there is only one subject S within the measurement range is “standing position (6.0 to 6.3)” → “end sitting position (4.1 to 4.3)”. It can be seen that the order is “→ post position (2.2 to 2.6)” → “exists outside measurement range (−2.0)”. This result is appropriate because it satisfies the property of “the risk of falling is high when the subject S is standing up and the risk of falling is low when the subject S is sleeping or absent”.

  The reason why the risk level of (f) has increased to 7.6 is that the risk level estimation system 1A mistaken the visitor (Another Man) for the patient. Such a phenomenon can be avoided by adjusting the measurement range and using time series data before and after.

Although (k) seems to be a standing posture, the risk is -2.0, which is the value when no person is present. In this risk level estimation system 1A, only the movement of the upper layer UL is seen outside the bed B when the risk level is estimated. In (k), since the subject person S leans against the desk, the height of the subject person S is detected lower than the actual height, and it is determined that the subject person S is absent. For these results, although the actual standing posture is regarded as absent, the absence is a posture after standing, so the standing posture is determined before the absence. There is no practical problem in estimating the degree.
Then, the result calculated by the risk level calculation means 24A of the risk level estimation unit 20A is output by the estimation result output unit 30A.

  As described above, the risk level estimation system 1A has a configuration in which the directly-upward image acquisition unit 10A includes the directly-upward image conversion unit 12A that converts the depth value image into the directly-upward image. Therefore, it is possible to acquire a depth value image from the head side of the bed B that hardly gives the subject S an unpleasant feeling. Further, as described in the description of the first embodiment, the risk level estimation using the image directly above can be performed with high accuracy. Therefore, by acquiring a depth value image from an oblique direction and converting the depth value image from the oblique direction into an image directly above, the risk level estimation system 1A is more accurate while using the depth value image from the oblique direction. Risk estimation can be performed. In addition, when acquiring a depth value image from an oblique direction by the directly above image acquiring unit 10A, the input unit 2 constituting the depth value image acquiring unit 11A can be installed as long as it is within a range that can be converted to an immediately above image. Therefore, the range in which the input unit 2 can be installed is greatly increased.

  Further, the risk level estimation system 1A includes a cube dividing unit 21A in which the determination layer dividing unit of the risk level estimation unit 20A divides into a plurality of cube regions. Therefore, it is possible to divide the region not only in the depth direction (Z-axis direction) but also in the planar region (X-axis direction, Y-axis direction), and perform risk estimation more accurately.

  In addition, by dividing into a plurality of cube regions by the cube dividing unit 21A, it is possible to estimate the risk level without using the position specifying unit 23A. For example, the section corresponding to the subject S can be determined, and the posture of the subject S can be estimated from the shape of the corresponding section. It is also possible to estimate the risk level of the subject person S by estimating the posture.

  The risk level estimation system 1A further includes a position specifying unit 23A for specifying the position of the object S and a risk level calculating unit 24A for calculating the risk level based on the specified position. It becomes the composition. Therefore, it is possible to estimate the degree of risk with higher accuracy. In the present embodiment, the position of the center of gravity of the subject S is used as the position specifying means 23A. Since it is possible to directly associate the position of the center of gravity with the posture of the subject person S, it is possible to estimate the degree of risk by estimating the posture of the subject person S from the position of the center of gravity.

  Further, since the risk level calculation means 24A is provided, for example, in the case of the risk level calculation method described above, the risk level can be calculated by changing the weight of the risk level by the subject S. In particular, the dangerous posture, the dangerous position, and the like vary depending on the symptom of the subject S, the shape of the bed B, the surrounding environment, and the like. Therefore, the risk level estimation unit 1A can perform risk level estimation suitable for the environment and the like.

  Further, the risk estimation target for performing the risk estimation in the present invention is not limited to a person like the target person S, but includes animals and objects other than humans, and is on the bed. Without being limited thereto, the present invention has a very wide range of application.

DESCRIPTION OF SYMBOLS 1, 1A ... Risk level estimation system 10, 10A ... Above image acquisition part 11, 11A ... Depth value image acquisition means 12A ... Just above image conversion means 20, 20A ... Risk level estimation part 21 ... Judgment layer division means 21A ... Cube division means 22, 22A ... layer of interest determination means 23A ... position specifying means 24A ... risk level calculation means 30, 30A ... estimation result output section

Claims (5)

An immediately-upward image acquisition unit configured to acquire an image directly above the risk estimation target formed based on the depth-value image formed by the depth value;
A risk estimation unit that estimates the risk based on the image directly above,
An estimation result output unit for outputting an estimation result by the risk estimation unit;
A risk estimation system comprising: The risk estimator is
A determination layer dividing means for dividing into determination layers for performing risk estimation using the depth value from the immediately above image;
Layer-of-interest determining means for determining a layer of interest from among the divided determination layers;
The risk level estimation system according to claim 1, wherein the risk level is estimated based on the target layer determined by the target layer determination means.   The risk estimation system according to claim 2, wherein the determination layer dividing unit includes a cube dividing unit that divides into a plurality of cube regions. The risk estimation unit further includes:
Position specifying means for specifying the position of the object;
A risk calculating means for calculating a risk based on the position specified by the position specifying means;
The risk level estimation system according to claim 2 or 3, characterized by comprising:   The risk level estimation system according to any one of claims 1 to 4, wherein the top image acquisition unit includes a top image conversion unit that converts the depth value image into the top image.