Automatic Alarm Communication Method for Home-based Elderly Care Based on Visual Monitoring

Abstract

This invention discloses an automatic alarm communication method for home-based elderly care based on visual monitoring, relating to the field of home visual monitoring technology. The method includes the following steps: converting home images to grayscale to obtain a grayscale image; obtaining a first filtering threshold and a second filtering threshold based on a first number of individual human images; binarizing the grayscale image based on the first and second grayscale thresholds to obtain a binarized image; obtaining a target contour based on the binarized image; constructing a real-time comparison value based on the target contour; obtaining a comparison threshold based on a second number of images of elderly people falling; generating a fall signal and automatically triggering an alarm based on the real-time comparison value and the comparison threshold. This invention addresses the problem in existing fall detection technologies that fail to further distinguish between squatting / bending and falls, leading to a decrease in the accuracy of fall detection when squatting / bending.

Description

Technical Field

[0001] This invention relates to the field of home-based visual monitoring technology, specifically to an automatic alarm communication method for home-based elderly care based on visual monitoring. Background Technology

[0002] Home-based elderly care has become the mainstream model of elderly care. However, elderly people who live alone, in empty nests, or are very old face many safety risks at home. Among them, accidental falls are the leading cause of serious injury, disability, or even death for the elderly. If they do not receive timely assistance after a fall, the consequences will be significantly worse. Therefore, it is necessary to conduct fall detection for the elderly.

[0003] Existing fall detection methods involve sensor detection and monitoring. Sensor detection requires real-time wearing, which may be inconvenient for elderly people with declining memory, as they may forget to wear the detection device. While existing monitoring-based detection uses model training to determine if a fall has occurred, the images of an elderly person squatting or bending over to pick something up are often similar to those of a fall, reducing the accuracy of fall detection. For example, patent application CN119723807A discloses a home-based elderly care visual smart monitoring alarm system, but this solution fails to further differentiate between images of squatting or bending over, leading to a decrease in fall detection accuracy. Existing fall detection technologies generally fail to further distinguish between squatting / bending over and falls, resulting in lower accuracy in fall detection during these situations. Summary of the Invention

[0004] This invention aims to at least partially solve one of the technical problems in the prior art by constructing a real-time comparison value based on a target contour; obtaining a comparison threshold based on images of a second number of elderly people falling; and generating a fall signal and automatically alarming based on the real-time comparison value and the comparison threshold. This addresses the problem in existing fall detection technologies that fail to further distinguish between squatting and bending over and falls, resulting in a decrease in the accuracy of fall detection when squatting and bending over.

[0005] To achieve the above objectives, this application provides an automatic alarm communication method for home-based elderly care based on visual monitoring, comprising the following steps:

[0006] A top-down view of the home environment is acquired in the white-hot mode of visual infrared monitoring and marked as a home image;

[0007] A grayscale image of a home is obtained by converting the image to grayscale.

[0008] A first screening threshold and a second screening threshold are obtained based on a first number of individual body images;

[0009] The grayscale image of the home is binarized based on the first grayscale threshold and the second grayscale threshold to obtain a binarized image of the home.

[0010] Obtain the target outline based on the home binarization image;

[0011] Real-time contrast values ​​are constructed based on the target contour;

[0012] A comparison threshold is obtained based on images of a second number of elderly people falling.

[0013] A fall signal is generated based on real-time comparison values ​​and comparison thresholds, and an alarm is automatically triggered.

[0014] Furthermore, the grayscale processing of the home image to obtain a home grayscale image includes the following sub-steps:

[0015] If the home image is a single-color-channel image, mark it as a home binarized image. If the home image is a red, green, and blue three-color-channel image, calculate the average of the red, green, and blue color channel values ​​of each pixel in the home image and merge them into a grayscale channel value to obtain a home grayscale image.

[0016] Furthermore, obtaining the first screening threshold and the second screening threshold based on the first number of individual body images includes the following sub-steps:

[0017] The first number of human images in the white-hot mode of the visual infrared monitoring are marked as human images; the human images are processed into grayscale to obtain a human grayscale image; the grayscale values ​​of the human parts in the human grayscale image are obtained and marked as human grayscale values.

[0018] Obtain the range of grayscale values ​​for the human body; divide the range of grayscale values ​​for the human body into a equal ranges, and mark them as the human body division ranges;

[0019] Count the frequency of each human body segmentation range and label it as the human body segmentation frequency;

[0020] The human body is divided into frequency groups and sorted in ascending order according to the median value of the human body division range to obtain the human body serial number, which is a positive integer starting from 1.

[0021] Furthermore, obtaining the first and second screening thresholds based on the first number of individual body images also includes the following sub-steps:

[0022] The frequency of human grayscale values ​​is obtained and denoted as Rp;

[0023] The human body segmentation threshold is calculated as: Rz = b × (Rp ÷ a); where Rz is the human body segmentation threshold and b is the ratio from 0 to 1;

[0024] The process involves checking if the frequency of a human segment is less than a threshold based on its sequence number from smallest to largest. If so, the segment is deleted, and the process continues until the frequency is greater than or equal to the threshold. Otherwise, the process stops. The process is repeated again, this time checking if the frequency of a human segment is less than a threshold based on its sequence number from largest to smallest. If so, the segment is deleted, and the process continues until the frequency is greater than or equal to the threshold. Otherwise, the process stops. Finally, the minimum and maximum values ​​of the remaining human segment frequencies are identified and marked as the first and second filtering thresholds, respectively.

[0025] Furthermore, the binarization of the grayscale image of the home, based on the first grayscale threshold and the second grayscale threshold, to obtain the binarized image of the home includes the following sub-steps:

[0026] Set grayscale values ​​in the home grayscale image that are greater than or equal to the first filtering threshold and less than or equal to the second filtering threshold to 0, and set grayscale values ​​in the home grayscale image that are less than the first filtering threshold and greater than the second filtering threshold to 255 to obtain a home binarized image.

[0027] Furthermore, obtaining the target contour based on the home binarized image includes the following sub-steps:

[0028] The binarized image of the home is scanned from top to bottom and left to right to obtain the first pixel whose grayscale value changes from 255 to 0, and this pixel is marked as the initial pixel. It is then checked whether any of the adjacent pixels around the initial pixel contain a pixel with a grayscale value of 0. If not, a new initial pixel is obtained until all adjacent pixels around the initial pixel are obtained, and this new initial pixel is marked as the starting pixel. Starting with the pixel adjacent to the starting pixel above it, and centering on the starting pixel, pixels are scanned clockwise to obtain the first pixel whose grayscale value changes from 255 to 0, and this pixel is marked as the search pixel. This process is repeated, starting with the pixel adjacent to the search pixel above it, and centering on the search pixel, scanning pixels clockwise to obtain new search pixels. The operation stops when a new search pixel matches the starting pixel. The starting pixel and all search pixels together form the target contour.

[0029] Furthermore, constructing real-time contrast values ​​based on the target contour includes the following sub-steps:

[0030] Establish a Cartesian coordinate system and label it as the constructed coordinate system. Place the binarized home image in the constructed coordinate system.

[0031] Rotate the binarized image of the home by 360° around its center, and obtain the maximum and minimum values ​​of the horizontal coordinate of a target contour in real time. Mark these values ​​as the maximum and minimum horizontal coordinates, respectively. Calculate the difference between the maximum and minimum horizontal coordinates and mark it as the difference.

[0032] The state of the binarized image at which the horizontal difference value is at its maximum during rotation is obtained and marked as the recognition state.

[0033] In recognition mode, the maximum and minimum values ​​of the vertical coordinate of a target contour are obtained and marked as the maximum and minimum vertical values, respectively; the difference between the maximum and minimum vertical values ​​is calculated and marked as the vertical difference value; the horizontal difference value is calculated and divided by the vertical difference value and marked as the real-time comparison value.

[0034] Furthermore, obtaining the comparison threshold based on the images of the second number of elderly people falling includes the following sub-steps:

[0035] Mark the second number of images of elderly people falling in the white-hot mode of visual infrared monitoring as fall images; obtain the real-time comparison value of the fall images and mark it as the fall comparison value;

[0036] Obtain the range of fall comparison values; divide the range of fall comparison values ​​into c equal ranges, and mark them as fall division ranges;

[0037] Count the frequency of each fall segment and label it as the fall segment frequency;

[0038] The frequency of falls is sorted in ascending order according to the median value of the fall range to obtain the fall sequence number, which is a positive integer starting from 1.

[0039] Furthermore, obtaining the comparison threshold based on the images of the second number of elderly people falling also includes the following sub-steps:

[0040] The frequency of fall comparison values ​​is recorded and denoted as Sp;

[0041] The fall classification threshold is calculated as: Sz = d × (Rp ÷ c); where Z is the fall classification threshold and d is the ratio from 0 to 1;

[0042] The system checks if the frequency of each fall segment is less than the fall segmentation threshold based on the fall sequence number from smallest to largest. If it is, the frequency of the fall segment is deleted, and the process continues until the frequency of the fall segment is greater than or equal to the fall segmentation threshold. If not, the process stops. The minimum value of the fall segmentation range corresponding to all remaining fall segment frequencies is obtained and marked as the comparison threshold.

[0043] Furthermore, generating a fall signal and automatically triggering an alarm based on real-time comparison values ​​and comparison thresholds includes the following sub-steps:

[0044] If the real-time comparison value is less than the comparison threshold, a normal behavior signal is generated; if the real-time comparison value is greater than or equal to the comparison threshold, a fall signal is generated and sent to the supervisor.

[0045] The beneficial effects of the present invention are as follows: The present invention constructs a real-time comparison value based on the target contour; obtains a comparison threshold based on images of a second number of elderly people falling; and generates a fall signal and automatically alarms based on the real-time comparison value and the comparison threshold. The advantage is that it can further distinguish between squatting and bending over and falling, and improves the accuracy of fall judgment in the case of squatting and bending over.

[0046] This invention constructs real-time contrast values ​​by using target contours. The advantage of this invention is that the constructed real-time contrast values ​​can be used to further distinguish between images of squatting and bending over and falling, thereby improving the accuracy of fall detection. Attached Figure Description

[0047] Figure 1 This is a flowchart illustrating the steps of the method of the present invention;

[0048] Figure 2 This is a schematic diagram illustrating the acquisition of search pixels according to the present invention;

[0049] Figure 3 This is a schematic diagram illustrating the acquisition of the identification status according to the present invention;

[0050] Figure 4 This is a schematic diagram of the horizontal maximum value, horizontal minimum value, vertical minimum value, and vertical maximum value of the present invention. Detailed Implementation

[0051] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0052] Example 1, please refer to Figure 1 As shown, this application provides an automatic alarm communication method for home-based elderly care based on visual monitoring, including the following steps:

[0053] Step S1: Obtain an overhead view image of the home environment in the white-hot mode of the visual infrared monitoring and mark it as a home image; because the human body has temperature, the infrared image in white-hot mode is used to facilitate human body identification.

[0054] Step S2 involves converting the home image to grayscale to obtain a grayscale image of the home. Step S2 includes the following sub-steps:

[0055] Step S201: If the home image is a single color channel image, mark the home image as a home binarized image; if the home image is a red, green and blue three color channel image, calculate the average of the red, green and blue three color channel values ​​of each pixel in the home image and merge them into a gray channel value to obtain a home grayscale image.

[0056] In practical applications, because the images acquired under the white-hot mode of different monitoring devices may be grayscale images or pseudo-color images, pseudo-color images are similar to grayscale images but are actually RGB images. Therefore, they need to be converted into grayscale images for convenient subsequent operations. For example, if the home image is an image with a single color channel, that is, a grayscale image, the home image will be marked as a home binarized image.

[0057] Step S3: Obtain a first screening threshold and a second screening threshold based on a first number of individual body images; Step S3 includes the following sub-steps:

[0058] Step S301: Mark the first number of human images in the white-hot mode of the visual infrared monitoring as human images; perform grayscale processing on the human images to obtain a human grayscale image; obtain the grayscale value of the human part in the human grayscale image and mark it as the human grayscale value; in order to obtain the range of human grayscale values, multiple human images are needed, and the more human images there are, the more accurate the range of human grayscale values ​​will be. Therefore, the larger the first number is set, the better. For example, the first number is set to 100.

[0059] Step S302: Obtain the range of grayscale values ​​of the human body; divide the range of grayscale values ​​of the human body into a equal ranges, and mark them as the human body division ranges; a is set to observe the number of grayscale values ​​of the human body in different ranges, since the grayscale values ​​are all integers, and at the same time ensure that the human body division ranges are integers, for example, a is 14.

[0060] Step S303: Count the frequency of each human body segmentation range and mark it as the human body segmentation frequency;

[0061] Step S304: Sort the human body segmentation frequency according to the median value of the human body segmentation range from small to large to obtain the human body sequence number. The human body sequence number is a positive integer starting from 1.

[0062] Step S305: Obtain the frequency of the grayscale value of the human body and label it as Rp;

[0063] Step S306, calculate the human body segmentation threshold as: Rz = b × (Rp ÷ a); where Rz is the human body segmentation threshold, and b is the ratio from 0 to 1; b is set to obtain the human body segmentation range with a small proportion of human body segmentation frequency, so b is small. At the same time, if Rp is larger, setting b to be smaller is more accurate, for example, b is 0.1.

[0064] Step S307: Determine if the frequency of human body segments is less than the human body segmentation threshold by sorting human body serial numbers from smallest to largest. If so, delete the human body segmentation frequency and continue the determination until the human body segmentation frequency is greater than or equal to the human body segmentation threshold. If not, stop the determination directly. Then, determine if the frequency of human body segments is less than the human body segmentation threshold by sorting human body serial numbers from largest to smallest. If so, delete the human body segmentation frequency and continue the determination until the human body segmentation frequency is greater than or equal to the human body segmentation threshold. If not, stop the determination directly. Obtain the minimum and maximum values ​​of the human body segmentation range corresponding to all remaining human body segmentation frequencies, and mark them as the first filtering threshold and the second filtering threshold, respectively. This process can filter out individual abnormal human body grayscale values ​​that are too large or too small, thereby obtaining an accurate human body grayscale value distribution range.

[0065] In practical applications, for example, in visual infrared monitoring, the detectable temperature image ranges from -50℃ to 50℃. The range of human grayscale values ​​is 178 to 220, which is divided into 14 equal ranges: 178 to 181, 181 to 184, ..., 217 to 220. The frequency of human grayscale values ​​is 12.21 million. Therefore, the human segmentation threshold is calculated as: Rz = 0.1 × (12.21 ÷ 14) = 8.7214, rounded to four decimal places. If the frequencies of human segmentation for 178 to 181, 181 to 184, ..., 214 to 217, and 217 to 220 are 6.1231, 12.1221, ..., 17.2398, and 4.1811 respectively, then 178 to 181 and 217 to 220 are deleted. The resulting first and second filtering thresholds are 181 and 217 respectively.

[0066] Step S4: Binarize the grayscale image of the home based on the first grayscale threshold and the second grayscale threshold to obtain a binarized image of the home; Step S4 includes the following sub-steps:

[0067] Step S401: Set the gray values ​​in the home grayscale image that are greater than or equal to the first filtering threshold and less than or equal to the second filtering threshold to 0, and set the gray values ​​in the home grayscale image that are less than the first filtering threshold and greater than the second filtering threshold to 255 to obtain a home binarized image; human body contours can be obtained.

[0068] In practical applications, gray values ​​greater than or equal to 181 and less than or equal to 217 in the home grayscale image are set to 0, and gray values ​​less than 181 and greater than or equal to 217 in the home grayscale image are set to 255 to obtain a home binarized image.

[0069] Step S5: Obtain the target contour based on the home binarized image; Step S5 includes the following sub-steps:

[0070] Step S501: Scan the binarized image from top to bottom and from left to right to obtain the first pixel whose grayscale value changes from 255 to 0, and mark it as the initial pixel. Determine whether the adjacent pixels around the initial pixel contain pixels with a grayscale value of 0. If not, obtain a new initial pixel until the adjacent pixels around the initial pixel are obtained, and mark this initial pixel as the starting pixel. Starting from the pixel adjacent to the starting pixel and centering on the starting pixel, scan the pixels in a clockwise direction to obtain the first pixel whose grayscale value changes from 255 to 0, and mark it as the search pixel. Repeat this process, starting from the pixel adjacent to the search pixel and centering on the search pixel, scanning the pixels in a clockwise direction to obtain new search pixels, until the new search pixel is the same as the starting pixel. Stop the operation when the new search pixel is the same as the starting pixel. The starting pixel and all the search pixels form the target contour.

[0071] For practical applications, please refer to Figure 2 As shown, this scanning method can filter out isolated interfering pixels and thus obtain the boundary contour of the human body.

[0072] Step S6: Construct real-time contrast values ​​based on the target contour; Step S6 includes the following sub-steps:

[0073] Step S601: Establish a Cartesian coordinate system and mark it as the construction coordinate system. Place the binarized home image in the construction coordinate system.

[0074] Step S602: Rotate the binarized image of the home by 360° around the center of the binarized image of the home, and obtain the maximum and minimum values ​​of the horizontal coordinate of a target contour in real time, and mark them as the horizontal maximum value and horizontal minimum value respectively. Calculate the difference between the horizontal maximum value and the horizontal minimum value, and mark it as the horizontal difference value.

[0075] Step S603: Obtain the state of the binarized image in the constructed coordinate system when the horizontal difference value is the largest during the rotation, and mark it as the recognition state;

[0076] Step S604: In the recognition state, obtain the maximum and minimum values ​​of the vertical coordinate of a target contour, and mark them as the maximum vertical value and the minimum vertical value, respectively; calculate the difference between the maximum vertical value and the minimum vertical value, and mark it as the vertical difference value; calculate the horizontal difference value and divide it by the vertical difference value, and mark it as the real-time comparison value; because the obtained image is a top view, when falling, the real-time comparison value is equivalent to the height being greater than the upper body width, while when a person squats or bends over, the corresponding value is the length less than the height being greater than the upper body width, and thus the size of the real-time comparison value can be used to distinguish whether a fall has occurred;

[0077] For practical applications, please refer to Figure 3 As shown, the process for obtaining the recognition status is as follows. Please refer to [link / reference]. Figure 4As shown, the maximum and minimum vertical values ​​are 182 and 108 respectively, and the maximum and minimum horizontal values ​​are 248 and 80 respectively. Therefore, the real-time comparison value is: (248-80)÷(182-108)=2.27.

[0078] Step S7: Obtain a comparison threshold based on the images of the second number of elderly people falling; Step S7 includes the following sub-steps:

[0079] Step S701: Mark the second number of images of the elderly falling in the white-hot mode of the visual infrared monitoring as fall images; obtain the real-time comparison value of the fall images and mark it as the fall comparison value; in order to obtain the range of the fall comparison value, multiple fall images are needed, and the more fall images there are, the more accurate the range of the fall comparison value will be. Therefore, the larger the second number is set, the better. For example, the second number is set to 200.

[0080] Step S702: Obtain the range of fall comparison values; divide the range of fall comparison values ​​into c equal ranges, and mark them as fall division ranges; c is set to the number of fall comparison values ​​distributed in different ranges, for example, c is 10;

[0081] Step S703: Count the frequency of each fall segment and mark it as the fall segment frequency;

[0082] Step S704: Sort the frequency of falls according to the median value of the fall division range from small to large to obtain the fall sequence number. The fall sequence number is a positive integer starting from 1.

[0083] Step S705: Obtain the frequency of the fall comparison value and label it as Sp;

[0084] Step S706, calculate the fall classification threshold as: Sz=d×(Rp÷c;where Z is the fall classification threshold, d is the proportion from 0 to 1; d is set to obtain the fall classification range with a small proportion of fall classification frequency, so b is small, for example b is 0.1.

[0085] Step S707: Determine whether the frequency of fall division is less than the fall division threshold according to the fall sequence number from smallest to largest. If so, delete the fall division frequency and continue to judge until the fall division frequency is greater than or equal to the fall division threshold and stop judging. If not, stop judging directly; obtain the minimum value of the fall division range corresponding to all remaining fall division frequencies and mark it as the comparison threshold.

[0086] In practical applications, for example, if the range of fall comparison values ​​is obtained from 1.8 to 3.8, and the frequency of fall comparison values ​​is taken as 200, the fall classification threshold is calculated as: Sz = 0.1 × (200 ÷ 10) = 2. If the fall classification frequency in the range of 1.8 to 2.0 is 1, then it is deleted; if the fall classification frequency in the range of 1.8 to 2.0 is 3, then the judgment stops, and the comparison threshold is 1.8.

[0087] Step S8: Generate a fall signal and automatically trigger an alarm based on the real-time comparison value and the comparison threshold; Step S8 includes the following sub-steps:

[0088] Step S801: If the real-time comparison value is less than the comparison threshold, a normal behavior signal is generated; if the real-time comparison value is greater than or equal to the comparison threshold, a fall signal is generated and sent to the supervisor. When falling, the real-time comparison value is equivalent to the height being greater than the upper body width, while when a person squats or bends over, it is equivalent to the length being less than the height being greater than the upper body width. Therefore, it can be determined that a fall has occurred if the real-time comparison value is greater than or equal to the comparison threshold.

[0089] In practical applications, if the real-time comparison value of 2.27 is greater than the comparison threshold of 1.8, a fall signal is generated and sent to the supervisor.

[0090] Example 2: This application also provides an electronic device, which may include: a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus. The memory stores computer-readable instructions, and the processor can call the instructions in the memory. When the computer-readable instructions are executed by the processor, the steps in the home-based elderly care automatic alarm communication method based on visual monitoring are performed to achieve the following functions: acquiring a top-down view image of the home environment in the white-hot mode of visual infrared monitoring and marking it as a home image; performing grayscale processing on the home image to obtain a home grayscale image; obtaining a first screening threshold and a second screening threshold based on a first number of human images; performing binarization processing on the home grayscale image based on the first grayscale threshold and the second grayscale threshold to obtain a home binarized image; obtaining a target contour based on the home binarized image; constructing a real-time comparison value based on the target contour; obtaining a comparison threshold based on images of a second number of elderly people falling; generating a fall signal and automatically alarming based on the real-time comparison value and the comparison threshold.

[0091] Furthermore, the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, and can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0092] Example 3: This application also provides a computer program product, which includes a computer program stored on a computer-readable storage medium. The computer program includes program instructions. When the program instructions are executed by a computer, the computer can execute the home-based elderly care automatic alarm communication method based on visual monitoring provided by the above methods. The method includes: acquiring a top-down view image of the home environment in a white-hot mode of visual infrared monitoring and marking it as a home image; performing grayscale processing on the home image to obtain a home grayscale image; acquiring a first filtering threshold and a second filtering threshold based on a first number of human images; performing binarization processing on the home grayscale image based on the first grayscale threshold and the second grayscale threshold to obtain a home binarized image; acquiring a target contour based on the home binarized image; constructing a real-time comparison value based on the target contour; acquiring a comparison threshold based on images of a second number of elderly people falling; generating a fall signal and automatically alarming based on the real-time comparison value and the comparison threshold.

[0093] Example 4: This application also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it performs the steps of the above-mentioned home-based elderly care automatic alarm communication method based on visual monitoring to achieve the following functions: acquiring a top-down view image of the home environment in the white-hot mode of visual infrared monitoring and marking it as a home image; performing grayscale processing on the home image to obtain a home grayscale image; obtaining a first filtering threshold and a second filtering threshold based on a first number of human images; performing binarization processing on the home grayscale image based on the first grayscale threshold and the second grayscale threshold to obtain a home binarized image; obtaining a target contour based on the home binarized image; constructing a real-time comparison value based on the target contour; obtaining a comparison threshold based on images of a second number of elderly people falling; generating a fall signal and automatically alarming based on the real-time comparison value and the comparison threshold.

[0094] Based on the above description of the embodiments, the embodiments of the present invention can be provided as methods, systems, or computer program products. Based on this understanding, the above technical solutions, in essence or in terms of their contribution to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or certain parts of the embodiments.

[0095] In the embodiments provided in this application, it should be understood that the disclosed system or method can be implemented in other ways. The embodiments described above are merely illustrative. For example, the division of modules or units is only a logical functional division, and there may be other division methods in actual implementation. Furthermore, multiple modules or units may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the coupling or direct coupling or communication connection shown or discussed may be through some communication interfaces. The indirect coupling or communication connection between systems, modules, and units may be electrical, mechanical, or other forms.

[0096] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A home-based elderly care automatic alarm communication method based on visual monitoring, characterized in that, Includes the following steps: A top-down view of the home environment is acquired in the white-hot mode of visual infrared monitoring and marked as a home image; A grayscale image of a home is obtained by converting the image to grayscale. A first screening threshold and a second screening threshold are obtained based on a first number of individual body images; The grayscale image of the home is binarized based on the first grayscale threshold and the second grayscale threshold to obtain a binarized image of the home. Obtain the target outline based on the home binarization image; Real-time contrast values ​​are constructed based on the target contour; A comparison threshold is obtained based on images of a second number of elderly people falling. A fall signal is generated based on real-time comparison values ​​and comparison thresholds, and an alarm is automatically triggered. Obtaining a grayscale image of a home by converting it to grayscale includes the following sub-steps: If the home image is a single color channel image, mark the home image as a home binarized image. If the home image is a red, green, and blue three color channel image, calculate the average of the red, green, and blue color channel values ​​of each pixel in the home image and merge them into a grayscale channel value to obtain a home grayscale image. Obtaining the first and second screening thresholds based on a first number of individual anatomy images includes the following sub-steps: The first number of human images in the white-heat mode of the visual infrared monitoring are marked as human images; the human images are then processed into grayscale to obtain human grayscale images; Obtain the grayscale values ​​of the human body portion in the grayscale image and label them as human body grayscale values; Obtain the range of grayscale values ​​for the human body; divide the range of grayscale values ​​for the human body into a equal ranges, and mark them as the human body division ranges; Count the frequency of each human body segmentation range and label it as the human body segmentation frequency; The human body is divided into frequency groups and sorted in ascending order according to the median value of the human body division range to obtain the human body serial number, which is a positive integer starting from 1.

2. The home-based elderly care automatic alarm communication method based on visual monitoring according to claim 1, characterized in that, The process of obtaining the first and second screening thresholds based on the first number of human images also includes the following sub-steps: The frequency of human grayscale values ​​is obtained and denoted as Rp; The human body segmentation threshold is calculated as: Rz = b × (Rp ÷ a); where Rz is the human body segmentation threshold and b is the ratio from 0 to 1; The process involves checking if the frequency of a human segment is less than a threshold based on its sequence number from smallest to largest. If so, the segment is deleted, and the process continues until the frequency is greater than or equal to the threshold. Otherwise, the process stops. The process is repeated again, this time checking if the frequency of a human segment is less than a threshold based on its sequence number from largest to smallest. If so, the segment is deleted, and the process continues until the frequency is greater than or equal to the threshold. Otherwise, the process stops. Finally, the minimum and maximum values ​​of the remaining human segment frequencies are identified and marked as the first and second filtering thresholds, respectively.

3. The home-based elderly care automatic alarm communication method based on visual monitoring according to claim 2, characterized in that, Binarizing the grayscale image of the home based on the first grayscale threshold and the second grayscale threshold to obtain the binarized image of the home includes the following sub-steps: Set grayscale values ​​in the home grayscale image that are greater than or equal to the first filtering threshold and less than or equal to the second filtering threshold to 0, and set grayscale values ​​in the home grayscale image that are less than the first filtering threshold and greater than the second filtering threshold to 255 to obtain a home binarized image.

4. The home-based elderly care automatic alarm communication method based on visual monitoring according to claim 3, characterized in that, Obtaining the target contour based on the home binarization image includes the following sub-steps: The binarized image of the home is scanned from top to bottom and left to right to obtain the first pixel whose grayscale value changes from 255 to 0, and this pixel is marked as the initial pixel. It is then checked whether any of the adjacent pixels around the initial pixel contain a pixel with a grayscale value of 0. If not, a new initial pixel is obtained until all adjacent pixels around the initial pixel are obtained, and this new initial pixel is marked as the starting pixel. Starting with the pixel adjacent to the starting pixel above it, and centering on the starting pixel, pixels are scanned clockwise to obtain the first pixel whose grayscale value changes from 255 to 0, and this pixel is marked as the search pixel. This process is repeated, starting with the pixel adjacent to the search pixel above it, and centering on the search pixel, scanning pixels clockwise to obtain new search pixels. The operation stops when a new search pixel matches the starting pixel. The starting pixel and all search pixels together form the target contour.

5. The home-based elderly care automatic alarm communication method based on visual monitoring according to claim 4, characterized in that, Constructing real-time contrast values ​​based on the target contour includes the following sub-steps: Establish a Cartesian coordinate system and label it as the constructed coordinate system. Place the binarized home image in the constructed coordinate system. Rotate the binarized image of the home by 360° around its center, and obtain the maximum and minimum values ​​of the horizontal coordinate of a target contour in real time. Mark these values ​​as the maximum and minimum horizontal coordinates, respectively. Calculate the difference between the maximum and minimum horizontal coordinates and mark it as the difference. The state of the binarized image at which the horizontal difference value is at its maximum during rotation is obtained and marked as the recognition state. In recognition mode, obtain the maximum and minimum values ​​of the ordinate of a target contour, and mark them as the maximum and minimum ordinate values, respectively. Calculate the difference between the maximum and minimum vertical values, and label it as the vertical difference value; The horizontal difference value is calculated and divided by the vertical difference value, and then marked as the real-time comparison value.

6. The home-based elderly care automatic alarm communication method based on visual monitoring according to claim 5, characterized in that, Obtaining the comparison threshold based on a second number of images of elderly people falling includes the following sub-steps: Mark the second number of images of elderly people falling in the white-hot mode of visual infrared monitoring as fall images; obtain the real-time comparison value of the fall images and mark it as the fall comparison value; Obtain the range of fall comparison values; divide the range of fall comparison values ​​into c equal ranges, and mark them as fall division ranges; Count the frequency of each fall segment and label it as the fall segment frequency; The frequency of falls is sorted in ascending order according to the median value of the fall range to obtain the fall sequence number, which is a positive integer starting from 1.

7. The home-based elderly care automatic alarm communication method based on visual monitoring according to claim 6, characterized in that, Obtaining the comparison threshold based on the second number of images of elderly people falling also includes the following sub-steps: The frequency of fall comparison values ​​is recorded and denoted as Sp; The fall classification threshold is calculated as: Sz = d × (Rp ÷ c); where Z is the fall classification threshold and d is the ratio from 0 to 1; The system checks if the frequency of each fall segment is less than the fall segmentation threshold based on the fall sequence number from smallest to largest. If it is, the frequency of the fall segment is deleted, and the process continues until the frequency of the fall segment is greater than or equal to the fall segmentation threshold. If not, the process stops. The minimum value of the fall segmentation range corresponding to all remaining fall segment frequencies is obtained and marked as the comparison threshold.

8. The home-based elderly care automatic alarm communication method based on visual monitoring according to claim 7, characterized in that, Generating a fall signal and automatically triggering an alarm based on real-time comparison values ​​and comparison thresholds includes the following sub-steps: If the real-time comparison value is less than the comparison threshold, a normal behavior signal is generated; if the real-time comparison value is greater than or equal to the comparison threshold, a fall signal is generated and sent to the supervisor.

Images