Method for monitoring a care recipient

Abstract

The application relates to the technical field of video image processing, and discloses a kind of automatic monitoring methods of care, the method comprises the following steps: obtaining the depth image data of care area, selecting image in depth image data, the phase difference calculation is carried out to the pixel depth value of image, and image change data is obtained;From image change data, the coordinate of the pixel depth value change is collected, the number of the coordinate of the pixel depth value change is calculated, the distance between the coordinate of the depth change value being positive and the coordinate of the depth change value being negative is calculated as movement distance;The number of the coordinate of the pixel depth value change is compared with the set change coordinate number threshold, the movement distance is compared with the set change distance threshold, and the behavior of the care object is judged.The application is different from the existing intelligent care mode, and positioning background or care object is not needed, the object directly monitored is moved, the calculation amount is smaller, and the risk of error is also reduced.

Description

Technical Field

[0001] This invention relates to the field of video image processing technology, and in particular to an automatic monitoring method for caregivers. Background Technology

[0002] With the expanding demand for care in hospitals, homes, and nursing homes, and the limitations of traditional human care such as its inability to provide 24-hour monitoring, intelligent care systems have gained significant development opportunities. Currently, most intelligent care systems require precise location of the background or the person being cared for.

[0003] For example, patent application CN112287821A discloses a method, device, computer equipment, and storage medium for monitoring the behavior of a caregiver. The method mainly includes acquiring depth data of a care scene captured by a depth camera; the depth camera taking pictures perpendicular to the bed in the care scene; the depth data including depth data at any position of the bed; determining the size and position of the bed based on the depth data; determining the position and depth value of the caregiver in the care scene based on the depth data, the size and position of the bed; and determining the behavior of the caregiver based on the position and depth value of the caregiver. This method determines the caregiver's behavior by locating the background environment and the caregiver, and then determining the caregiver's positional relationship with the background environment.

[0004] For example, patent application CN105868707A discloses a real-time bed fall detection method based on depth image information, including: (1) acquiring a depth image of an indoor scene using a depth sensor; (2) updating the tracking area in the depth image; (3) extracting the eight-neighbor difference features of each pixel in the tracking area obtained in step (2) by continuously changing the offset scale; (4) acquiring the head area; (5) locating the head center; (6) acquiring the upper body area; (7) optimizing head positioning; (8) human body confirmation; (9) extracting height features; (10) bed fall detection: using a trained bed fall classifier to classify and detect the above height feature vectors to obtain the bed fall detection result. This method directly and accurately locates the body parts of the person being cared for, achieving the purpose of detecting whether the person being cared for has fallen off the bed.

[0005] Both of the above methods can accurately monitor the time of care, but they require a lot of data to be detected and the analysis process is more complex. Not only is the amount of calculation large, but most of them need to rely on cloud-based auxiliary calculations. Moreover, the risk of errors actually increases with the number of analysis steps. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide an automatic monitoring method for caregiving, which can more easily meet caregiving needs.

[0007] The automatic monitoring method for caregiving disclosed in this invention includes the following steps:

[0008] Acquire depth image data of the care area, select an image from the depth image data, calculate the phase difference between the corresponding pixel depth values ​​of the image, and obtain image change data;

[0009] Collect coordinates of pixel depth values ​​that change from image change data, calculate the number of coordinates of pixel depth values ​​that change, and separately count the coordinates of positive depth change values ​​and the coordinates of negative depth change values. Calculate the distance between the coordinates of positive depth change values ​​and the coordinates of negative depth change values ​​as the motion distance.

[0010] The behavior of the monitored object is determined by comparing the number of coordinates where the pixel depth value changes with a set threshold for the number of coordinate changes, and by comparing the movement distance with a set threshold for the distance change.

[0011] Preferably, when extracting images from depth image data, the nth frame image and the ntth frame image are selected, and the pixel depth values ​​of the nth frame image and the ntth frame image are compared to obtain image change data, where n is a positive integer greater than t and t is a positive integer greater than 1.

[0012] Preferably, the threshold for the number of changed coordinates includes a first quantity alarm threshold, and the threshold for the changed distance includes a first distance alarm threshold;

[0013] When the number of coordinates where the pixel depth value changes exceeds the first quantity alarm threshold, and the movement distance exceeds the first distance alarm threshold, it is determined that the object under care has engaged in Level 1 manual care behavior, and a Level 1 alarm signal is issued.

[0014] Preferably, the first quantity alarm threshold is the number of coordinates of pixel depth value changes caused by 2 / 3 movement of the human body in a bedridden state, and the first distance alarm threshold is 1 / 2 of the width of the bed in the care area.

[0015] Preferably, the threshold for the number of changed coordinates includes a second quantity alarm threshold, and the threshold for the changed distance includes a second distance alarm threshold;

[0016] When the number of coordinates where the pixel depth value changes is greater than the second alarm threshold and less than the first alarm threshold, and the movement distance is greater than the second alarm threshold and less than the first alarm threshold, it is determined that the person being cared for has engaged in level two human care behavior, and a level two alarm signal is issued.

[0017] Preferably, the second quantity alarm threshold is the number of coordinates of pixel depth value changes caused by 1 / 3 movement of the human body in a bedridden state, and the second distance alarm threshold is 1 / 3 of the width of the bed in the care area.

[0018] Preferably, when calculating the difference between the pixel depth values ​​of two frames, the pixel depth value of the image with the later time is subtracted from the pixel depth value of the image with the earlier time to obtain the image change data;

[0019] The region with a negative depth change value in the image change data is taken as the starting point of the moving object, and the region with a positive depth change value in the image change data is taken as the ending point of the moving object. The distance between the starting point and the ending point is the motion distance.

[0020] Preferably, when calculating the movement distance, the coordinates of the center point of the region with a positive depth change value and the coordinates of the center point of the region with a negative depth change value are calculated, and the distance between the center point coordinates with a positive depth change value and the center point coordinates with a negative depth change value is taken as the movement distance.

[0021] Preferably, the following calculation method is used when calculating the coordinate center point with a positive depth change value and the coordinate center point with a negative depth change value:

[0022] The coordinates of the center point of the coordinate system with a positive depth change value are marked as (P). x P y The coordinates of the center point with a positive depth change value are marked as (N). x N y ),

[0023] ;

[0024] ;

[0025] This represents the number of coordinates whose depth change value is positive. This represents the number of coordinates with negative depth change values. This represents a set of coordinates whose depth changes are positive. A set of coordinates representing a negative depth change.

[0026] The beneficial effects of this invention are as follows: This application differs from existing intelligent care methods in that it does not require locating the background or the object being cared for. It directly monitors the moving object and determines the behavior of the object being cared for based on the number of coordinates and distances where the pixel depth value changes. Compared with existing intelligent care methods, it requires less computation, which helps to reduce equipment investment costs. It also reduces the number of analysis steps and the risk of errors. Attached Figure Description

[0027] Figure 1 This is a flowchart of the automatic monitoring method for caregiving described in this application;

[0028] Figure 2 This is a schematic diagram illustrating the phase differences between each frame of the image;

[0029] Figure 3a This is a schematic diagram of the image coordinates of the nth frame;

[0030] Figure 3b This is a schematic diagram of the image coordinates of the nt-th frame;

[0031] Figure 3c This is a schematic diagram of the image coordinates of the image change data. Detailed Implementation

[0032] The present invention will be further described below.

[0033] The automatic monitoring method for caregivers described in this application is typically used in hospital wards or homes, and enables automatic monitoring of the caregiver when the caregiver is not present. When the caregiver is not present, the overall environment of the care area is relatively simple and quiet, with no other moving objects besides the caregiver, which provides the basic conditions for the implementation of this application. The automatic monitoring method for caregivers described in this application includes the following steps:

[0034] Acquire depth image data of the care area, select an image from the depth image data, calculate the phase difference between the corresponding pixel depth values ​​of the image, and obtain image change data;

[0035] Collect coordinates of pixel depth values ​​that change from image change data, calculate the number of coordinates of pixel depth values ​​that change, and separately count the coordinates of positive depth change values ​​and the coordinates of negative depth change values. Calculate the distance between the coordinates of positive depth change values ​​and the coordinates of negative depth change values ​​as the distance of pixel depth value change.

[0036] The behavior of the monitored object is determined by comparing the number of coordinates where the pixel depth value changes with a set threshold for the number of coordinate changes, and by comparing the distance at which the pixel depth value changes with a set threshold for the distance at which the change occurs.

[0037] Depth image data can typically be acquired using existing equipment such as depth cameras. Depth images, also known as distance images, differ from ordinary image data in that they contain depth information for each pixel. By calculating the phase difference between two frames of depth image data, image change data can be obtained. Analyzing this data reveals how the monitored area changed within the time frame between the two frames. This method is similar to the inter-frame differencing method used in existing technologies. (See figure.) Figure 3a and Figure 3b These represent two images extracted from depth image data, where depth values ​​are represented numerically. Taking a depth camera as an example, a smaller number indicates a closer distance to the depth camera. Figure 3b and Figure 3a By subtracting the corresponding values, we can obtain the image change data. Figure 3c . Figure 3c The coordinates of the number 0 indicate that the pixel depth value has not changed, while positive or negative numbers indicate that the depth value has changed. From this image change data, the coordinates of changed pixel depth values ​​can be directly counted. The more coordinates of changed pixel depth values, the greater the range of motion of the person being cared for. For example, if the person is only moving their hand, the number of coordinates of changed pixel depth values ​​will be small, while the number of coordinates of changed pixel depth values ​​will be large if the entire body is moving. However, judging the behavior of the person being cared for solely based on the number of coordinates of changed pixel depth values ​​may lead to misjudgment. For example, if the person being cared for turns over or changes their bed posture, the number of coordinates of changed pixel depth values ​​will be large, but no special care is needed. Therefore, it is also necessary to supplement this by considering the distance of the person's movement. Figure 3c The regions containing positive or negative values ​​represent the object's position before and after movement. The distance between coordinates showing positive depth changes and coordinates showing negative depth changes can be used as the movement distance of the monitored object. By comparing the number of coordinates showing changes in pixel depth values ​​and the movement distance, the behavior of the monitored object can be determined, and feedback can be provided to the caregiver. Thus, with real-time monitoring, image extraction, and statistical analysis, the goal of automatic monitoring can be achieved.

[0038] Obtaining image change data can be done using the existing inter-frame difference method, which involves performing a phase difference operation on two adjacent frames. However, this application directly judges the behavior of the person being cared for based on motion. Sufficient image differences from the person being cared for are required for accurate judgment. For bedridden patients, movements are usually slow, and except in special cases such as falls from bed, calculating the difference between two adjacent frames is insufficient for monitoring purposes. Therefore, in a preferred embodiment of this application, when extracting images from depth image data, the nth frame and the ntth frame are selected, and the pixel depth values ​​of the nth and ntth frames are calculated to obtain image change data. Here, n is a positive integer greater than t, and t is a positive integer greater than 1. For example, when t equals 4, the 5th frame is compared with the 1st frame, the 6th frame with the 2nd frame, and so on, enabling real-time monitoring. The specific value of t can be determined based on the frame rate of the acquired depth image data and the person being cared for.

[0039] As mentioned above, this application compares the number of coordinates where pixel depth values ​​change with a threshold for the number of changed coordinates, and the movement distance with a set threshold for the change distance, to determine the magnitude and distance of the movement of the person being cared for, thereby determining the behavior of the person being cared for. In a preferred embodiment of this application, the threshold for the number of changed coordinates includes a first quantity alarm threshold, and the threshold for the change distance includes a first distance alarm threshold. When the number of coordinates where pixel depth values ​​change is greater than the first quantity alarm threshold, and the movement distance is greater than the first distance alarm threshold, it is determined that the person being cared for has engaged in Level 1 manual care behavior, and a Level 1 alarm signal is issued. The Level 1 manual care behavior is set according to needs. For example, the first quantity alarm threshold can be determined based on the number of coordinates where pixel depth values ​​change when the person falls out of bed, gets up, or gets out of bed, and the first distance alarm threshold is also determined based on the movement distance when the person falls out of bed, gets up, or gets out of bed, for example, in a preferred embodiment of this application, the first quantity alarm threshold is the number of coordinates where pixel depth values ​​change when the person moves 2 / 3 of the body while in a bedridden state, and the first distance alarm threshold is 1 / 2 of the width of the bed in the care area. The Level 1 alarm signal can be sent to caregivers via local area network or the Internet. After receiving the Level 1 alarm signal, caregivers can check the status of the caregivers through monitoring. If it is confirmed that a caregiver needs help, they will rush to the scene to handle the situation.

[0040] Different alarm levels can be set for different caregivers. For example, a level 1 alarm can be set for caregivers in relatively good condition, while multiple alarm levels can be set for more critical caregivers. Specifically, in the preferred embodiment of this application, the threshold for the number of changing coordinates includes a second alarm threshold, and the threshold for the change in distance includes a second alarm threshold. When the number of coordinates where the pixel depth value changes is greater than the second alarm threshold and less than the first alarm threshold, and the movement distance is greater than the second alarm threshold and less than the first alarm threshold, it is determined that the caregiver has engaged in a level 2 human intervention behavior, and an alarm signal is issued. The probability of danger for level 2 human intervention behavior is less than that for level 1 human intervention behavior. Therefore, it can be used to automatically monitor behaviors such as significant turning over or sitting up in caregivers. In the preferred embodiment of this application, the second alarm threshold is the number of coordinates where the pixel depth value changes due to 1 / 3 of the human body movement while lying in bed, and the second alarm threshold is 1 / 3 of the width of the bed in the care area. Here, the human body can usually be an average human body type, or multiple options can be set according to human height, and different body type options can be selected according to different caregivers. Upon receiving a Level 2 alarm signal, caregivers can also monitor the condition of those under their care. If it is confirmed that someone needs assistance, they can then rush to the scene to handle the situation.

[0041] When calculating the difference between the pixel depth values ​​of two images, one can subtract the pixel depth value of the earlier image from the pixel depth value of the later image, or vice versa. The preferred embodiment of this application uses the former subtraction method. In the image change data obtained using the former method, regions with negative depth change values ​​can be considered as the starting point of the moving object, and regions with positive depth change values ​​can be considered as the ending point. The distance between the starting point and the ending point is the movement distance. If the latter subtraction method is used, the method for determining the starting point and the ending point is exactly the opposite.

[0042] Since the negative coordinate region and the positive coordinate region are a range, in order to accurately calculate the movement distance, when calculating the movement distance, the coordinates of the center point of the region with a positive depth change value and the coordinates of the center point of the region with a negative depth change value are calculated, and the distance between the center point coordinates of the positive depth change value and the center point coordinates of the negative depth change value is taken as the movement distance.

[0043] Regarding the specific calculation method of this application, as follows: Figures 3a-3c In the calculation, the following can be defined: ,in, For the first frame down The depth value of a pixel. For the first frame down The depth value of a pixel. Pixel number Frame and the Frame differences.

[0044] Collect a set of coordinates with positive depth change values ​​from image change data: Collect the set of coordinates with negative depth change values ​​from the image change data: The number of coordinates with positive depth change values ​​was counted. And the number of coordinates with negative depth change values. .For example Figure 3c middle, , , , .

[0045] The coordinates of the center point of the coordinate system with a positive depth change value are marked as (P). x P y The coordinates of the center point with a positive depth change value are marked as (N). x N y ),

[0046] ;

[0047] ;

[0048] Based on the above calculations Figure 3c The coordinates of the center point of the coordinate system with a positive depth change value are: The coordinates of the center point where the depth change value is negative are: The distance between the two is 2. Of course, this embodiment is an example for ease of explanation, and the numbers in the figure do not represent actual pixels.

Claims

1. An automatic monitoring method for caregivers, characterized in that, Includes the following steps: Acquire depth image data of the care area, select an image from the depth image data, calculate the phase difference between the corresponding pixel depth values ​​of the image, and obtain image change data; Collect coordinates of pixel depth values ​​that change from image change data, calculate the number of coordinates of pixel depth values ​​that change, and separately count the coordinates of positive depth change values ​​and the coordinates of negative depth change values. Calculate the distance between the coordinates of positive depth change values ​​and the coordinates of negative depth change values ​​as the motion distance. The number of coordinates where pixel depth values ​​change is compared with a set threshold for the number of coordinate changes, and the movement distance is compared with a set threshold for the distance change to determine the behavior of the person being cared for. The threshold for the number of changing coordinates includes a first quantity alarm threshold, and the threshold for the changing distance includes a first distance alarm threshold; When the number of coordinates where the pixel depth value changes is greater than the first quantity alarm threshold, and the movement distance is greater than the first distance alarm threshold, it is determined that the object under care has engaged in Level 1 manual care behavior, and a Level 1 alarm signal is issued. The first quantity alarm threshold is the number of coordinates of pixel depth value changes caused by 2 / 3 movement of the human body in a bedridden state, and the first distance alarm threshold is 1 / 2 of the width of the bed in the care area.

2. The automatic monitoring method for caregiving as described in claim 1, characterized in that, When extracting images from depth image data, the nth frame image and the ntth frame image are selected, and the pixel depth values ​​of the nth frame image and the ntth frame image are compared to calculate the difference to obtain image change data, where n is a positive integer greater than t and t is a positive integer greater than 1.

3. The automatic monitoring method for caregiving as described in claim 1, characterized in that, The threshold for the number of changed coordinates includes a second quantity alarm threshold, and the threshold for the change in distance includes a second distance alarm threshold; When the number of coordinates where the pixel depth value changes is greater than the second alarm threshold and less than the first alarm threshold, and the movement distance is greater than the second alarm threshold and less than the first alarm threshold, it is determined that the person being cared for has engaged in level two human care behavior, and a level two alarm signal is issued.

4. The automatic monitoring method for caregiving as described in claim 3, characterized in that, The second quantity alarm threshold is the number of coordinates of pixel depth value changes caused by 1 / 3 movement of the human body in a bedridden state, and the second distance alarm threshold is 1 / 3 of the width of the bed in the care area.

5. The automatic monitoring method for caregiving as described in claim 1, characterized in that, When calculating the difference between the pixel depth values ​​of two frames, the pixel depth value of the image with the later time is subtracted from the pixel depth value of the image with the earlier time to obtain the image change data. The region with a negative depth change value in the image change data is taken as the starting point of the moving object, and the region with a positive depth change value in the image change data is taken as the ending point of the moving object. The distance between the starting point and the ending point is the motion distance.

6. The automatic monitoring method for caregiving as described in claim 1 or 5, characterized in that, When calculating the movement distance, the coordinates of the center point of the region with a positive depth change value and the coordinates of the center point of the region with a negative depth change value are calculated, and the distance between the center point coordinates with a positive depth change value and the center point coordinates with a negative depth change value is taken as the movement distance.

7. The automatic monitoring method for caregiving as described in claim 6, characterized in that, The following calculation method is used when calculating the coordinate center point with a positive depth change value and the coordinate center point with a negative depth change value: The coordinates of the center point of the coordinate system with a positive depth change value are marked as (P). x P y The coordinates of the center point with a positive depth change value are marked as (N). x N y ), ； ； This represents the number of coordinates whose depth change value is positive. This represents the number of coordinates with negative depth change values. This represents a set of coordinates whose depth changes are positive. A set of coordinates representing a negative depth change.

Images