A facial expression recognition method based on intelligent nursing and nursing equipment thereof

Abstract

The present invention discloses a facial expression recognition method based on intelligent nursing and nursing equipment thereof, which relates to the field of intelligent nursing technology; in order to solve the problem that the judgment of the patient's condition is not accurate enough and the scope of application is narrow; the recognition method includes the following steps: collecting the static facial image of the person who needs to be cared for in advance; preprocessing the collected static facial image; processing the preprocessed facial image into equal blocks, and extracting the texture feature points in the facial image using the LBP algorithm; deforming the texture feature points according to the facial animation parameter rules obtained by statistical learning and in combination with Bezier surface control; the nursing equipment of the recognition method includes a nursing mechanism for real-time monitoring of the person being cared for, and the nursing mechanism includes a bed body and a monitoring part fixed to the top of two or more legs. The present invention has a wide scope of application and high accuracy, effectively saves the energy and time of medical staff in caring, and at the same time ensures the life safety of the cared person.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent nursing, and in particular to a facial expression recognition method based on intelligent nursing and nursing equipment thereof. Background Art

[0002] With the rapid development of medical technology, medical workers and researchers have been committed to protecting our health and conquering many stubborn diseases. However, while ensuring high standards of clinical medical care, it is becoming increasingly important for hospitals to provide patients with a safe, caring, thoughtful and humane ward care environment and conditions. A safe and efficient ward care system has gradually become a new standard for measuring the quality of hospital care, especially in intensive care units and wards for patients who are recovering or have difficulty walking. If a patient has an abnormal reaction or a serious complication and there is no one to take care of the ward, the patient may lose the best time for treatment.

[0003] After searching, the Chinese patent application number CN201410826544.3 discloses a critically ill patient monitoring system based on intelligent video, including two image acquisition modules, a medical staff identification module, a patient monitoring module, a human-computer interaction module, an alarm module, and a data storage module; the first image acquisition module is installed in the intensive care unit and can continuously acquire the overall image of the bed and the patient. The critically ill patient monitoring system based on intelligent video in the above patent has the following shortcomings: it only detects dangerous movements of the patient's limbs, the judgment basis is not accurate enough, it is not suitable for patients with severe illness or limbs that cannot move, and the scope of application is relatively narrow. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a facial expression recognition method based on intelligent care and a nursing device thereof.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] A facial expression recognition method based on intelligent care includes the following steps:

[0007] S1: Collect static facial images of the person who needs care in advance;

[0008] S2: preprocessing the collected facial static images;

[0009] S3: The pre-processed facial image is divided into equal blocks and the texture feature points in the facial image are extracted using the LBP algorithm;

[0010] S4: According to the facial animation parameter rules obtained by statistical learning, the texture feature points are deformed with Bezier surface control to simulate and generate dynamic facial expression images;

[0011] S5: Packaging and transmitting the obtained single-category image data to the nursing information database for storage;

[0012] S6: Using the care institution to monitor the cared person in real time, capturing the face range based on the center of both eyes and capturing the face image;

[0013] S7: extracting LBP expression feature values ​​of the facial image through the LBP algorithm, comparing the extracted LBP expression feature values ​​with the parameter values ​​in the nursing information database, and obtaining the facial expression recognition result;

[0014] S8: Automatically determine whether the expression recognition result causes an alarm based on the warning mark in the care information database;

[0015] The facial static image acquisition method includes picture input, real-time photography, and video capture;

[0016] The preprocessing includes size normalization processing, noise reduction processing and grayscale processing methods, and the unit of size normalization processing is pixel;

[0017] The noise reduction process is one of linear mean filtering and Gaussian filtering.

[0018] Preferably, the method for equally partitioned block processing comprises the following steps:

[0019] S101: Divide the detected facial image into small blocks of n×n pixel units;

[0020] S102: Calculate the LBP value in each small block, and connect the LBP values ​​in all small blocks into a feature vector to obtain the LBP texture feature vector of the detected facial image.

[0021] Preferably, the generating of dynamic facial expression images specifically comprises the following steps:

[0022] S301: using the Bezier surface control deformation method and facial animation parameter rules, generating a golden skull mesh from texture feature points extracted from the facial image;

[0023] S302: Mapping the preprocessed facial image as a texture to the golden skull mesh, and controlling the expression change by using the image's gradient domain information and parameter mapping method;

[0024] S303: Using Open GL rendering, a facial image with various basic expressions is obtained.

[0025] Preferably, the various basic expressions include a safe expression, a warning expression and a dangerous expression, and the safe expression, the warning expression and the dangerous expression all include micro-expressions with different ranges of facial changes.

[0026] Preferably, the single type of image data is static facial images and various dynamic facial expression images derived from static facial images; the nursing information database also includes a deletion module for deleting expired caregiver image information.

[0027] Preferably, the step of comparing the extracted LBP expression feature values ​​with the parameter values ​​in the nursing information database specifically comprises the following steps:

[0028] S401: Compare the LBP expression feature value with the feature values ​​of various basic expressions, and confirm that the LBP value repetition point is greater than the set threshold, and the facial image of the caregiver with the most LBP value repetition points;

[0029] S402: According to the dynamic facial expression image, confirm the expression classification of the extracted LBP expression feature value.

[0030] Preferably: the warning mark includes a warning level for distinguishing alarm information and an editing block for medical staff to input micro-expressions that each caregiver needs to be reminded of; the warning levels include a first category of no alarm, a second category of selective alarm, and a third category of direct alarm, which correspond to safety expressions, warning expressions, and danger expressions, respectively.

[0031] A nursing device based on a facial expression recognition method for intelligent nursing, comprising a nursing mechanism for real-time monitoring of a person being cared for, the nursing mechanism comprising a bed body fixed to the top ends of two or more legs and a monitoring unit, the top outer wall of the bed body being fixedly connected to a lower bed board, an upper bed board being arranged on one side edge of the lower bed board; a top support piece being fixedly connected to the bottom inner wall of the bed body, an extended end of the top support piece and the bottom outer wall of the upper bed board being rotatably matched through a connecting portion, the connecting portion being composed of an arc-shaped plate fixedly connected to the extended end of the top support piece and a concave seat fixedly connected to the bottom outer wall of the upper bed board; the monitoring unit being connected by a support The invention comprises a shooting head A which is fixed to one end of the bed body by a column, and a shooting head B which is fixed to one end of the lower bed board by a bracket; the interior of the bed body is composed of a cleaning part and a collecting part; an inner wall on one side of the lower bed board is provided with an automatic opening and closing part of the box door, and the automatic opening and closing part of the box door comprises a fixed plate, a motor and a transmission gear, the outer wall of the fixed plate is slidably connected to the inner wall of the lower bed board, and the outer wall on one side of the fixed plate is fixedly connected with a rack; the bottom outer wall of the motor is fixedly connected to the inner wall of the bed body near the top; the bottom end of the transmission gear is connected to the output end of the motor through a connecting shaft, and the outer wall of the transmission gear is meshed with the outer wall of the rack.

[0032] Preferably: the cleaning part includes a water tank, a flushing head and an inclined cleaning head, the bottom outer wall of the water tank is arranged on the bottom inner wall of the bed; one end of the flushing head is fixedly connected to the top inner wall of the bed, and the input end of the flushing head is connected to the outer wall of one side of the water tank through a water pipe; one end of the cleaning head is fixedly connected to the inner wall of one side of the bed near the top, and the input end of the cleaning head is connected to the outer wall of one side of the water tank through a conduit.

[0033] Preferably: the collecting part includes a receiving box, a waste storage box, an inlet, a baffle plate and a weight reminder structure, one side outer wall of the receiving box is fixed to the one side outer wall of the fixing plate through a connecting block; the top of the waste storage box is arranged on the bottom outer wall of the bed; the inlet is arranged on the bottom outer wall of the bed, and the inner wall on the opposite side of the inlet is provided with a slide rail; the outer walls on both sides of the baffle plate are slidably connected to the inner walls of the two slide rails through sliders, and the top outer wall of the baffle plate close to one edge is connected to the bottom outer wall of the receiving box through a supporting block; the weight reminder structure is composed of a containing group and a detection group.

[0034] The beneficial effects of the present invention are:

[0035] 1. When providing care, the present invention calculates the expression feature value of the patient based on the real-time captured image, searches for and calculates the facial static image that fits the expression feature value in the care information database, thereby confirming the identity information of the person being cared for, and confirms the expression classification of the extracted expression feature value based on the facial images of various basic expressions derived from the facial static image, thereby analyzing the content of the patient's expression, thereby obtaining the actual situation of the patient, with a wide range of applications and high accuracy, effectively saving the energy and time of medical staff in caring for the patient, while ensuring the life safety of the cared person.

[0036] 2. The present invention pre-processes the static facial image of the caregiver collected by static or dynamic methods to improve the clarity; the pre-processed facial image is divided into equal blocks, and the texture feature points in the facial image are extracted using the LBP algorithm to ensure the accuracy of subsequent face recognition; the texture feature points extracted from the facial image are generated into vivid images with various basic expressions using the Bezier surface control deformation method and facial animation parameter rules to ensure the timeliness and accuracy of monitoring and recognition results.

[0037] 3. The present invention uses the warning level of the monitored facial expression and the micro-expressions of each caregiver that need to be reminded input by the medical staff as the basis for judgment, and performs high-quality monitoring according to the actual situation of each caregiver, so as to provide patients with better services.

[0038] 4. The present invention sets structures such as the automatic opening and closing part of the box door, starts the motor to drive the transmission gear to rotate, and drives the fixed plate to slide back and forth in the lower bed plate through the meshing of the rack, so that the discharge port is opened / closed, and the fixed plate drives the receiving box to move synchronously through the connecting block. At the same time, the receiving box drives the barrier plate to move therewith through the supporting block, so as to synchronously complete the opening / closing of the discharge port and the closing / opening of the inlet through the barrier plate, which is convenient for collecting excrement and preventing the odor in the waste storage box from being emitted.

[0039] When the loading box stops moving, the loading plate is completely out of the wrapping range of the loading box, and the feeding port is adjusted to be closed and the loading port is opened.

[0040] 6. The present invention provides a detection group and other structures. After the carrying plate enters the wrapping range of the receiving box, as the amount of excrement carried during use increases, the weight pressure causes the carrying plate to move downward as a whole. During the downward movement, the first group of pressure-bearing components, the second group of pressure-bearing components and the third group of pressure-bearing components arranged vertically are squeezed in sequence. Specifically, when the carrying plate moves downward, the sphere on the side is squeezed to shrink inward, thereby causing the elastic member to contract, and the T-shaped rod slides through the partition plate, so that one end surface of the T-shaped rod contacts the collision sensor directly in front of it in the horizontal direction. When the carrying plate indirectly touches the collision sensors on the first group of pressure-bearing components and the second group of pressure-bearing components, the control module determines whether to close the automatic opening and closing part of the door within XX time according to the facial image captured by the camera B. When the carrying plate indirectly touches the collision sensor on the third group of pressure-bearing components, the control module determines that the automatic opening and closing part of the door is still not controlled to be closed according to the facial image captured by the camera B, and the control module initiates three types of direct alarms to prompt medical staff to arrive at the scene in time to reduce safety hazards. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 A schematic diagram of a process of a facial expression recognition method based on intelligent care proposed by the present invention;

[0042] Figure 2This is a left-view structural schematic diagram of a nursing device for a facial expression recognition method based on intelligent nursing proposed by the present invention;

[0043] Figure 3 A right-view partial cross-sectional structural schematic diagram of a nursing device for a facial expression recognition method based on intelligent nursing proposed by the present invention;

[0044] Figure 4 This is a structural schematic diagram of the connection state of a fixing plate and a receiving box of a nursing device of a facial expression recognition method based on intelligent nursing proposed by the present invention;

[0045] Figure 5 A schematic diagram of a cross-sectional structure of a receiving box of a nursing device based on a facial expression recognition method for intelligent nursing proposed by the present invention when viewed from above;

[0046] Figure 6 A schematic diagram of the side cross-sectional structure of a receiving box of a nursing device for a facial expression recognition method based on intelligent nursing proposed by the present invention;

[0047] Figure 7 A schematic diagram of a partial cross-sectional structure of a bed body in a side view of a nursing device for a facial expression recognition method based on intelligent nursing proposed by the present invention;

[0048] Figure 8 This is a circuit flow diagram of a nursing device for a facial expression recognition method based on intelligent care proposed by the present invention.

[0049] In the figure: 1 bed body, 2 top support, 3 shooting head A, 4 upper bed board, 5 fixing board, 6 shooting head B, 7 protective pad, 8 lower bed board, 9 legs, 10 storage box, 11 waste storage box, 12 motor, 13 receiving box, 14 flushing head, 15 inlet, 16 elastic member, 17 elastic cover, 18 water tank, 19 observation window, 20 water inlet, 21 transmission gear, 22 guide baffle, 23 adsorption layer, 24 barrier plate, 25 cleaning head, 26 T-bar, 27 collision sensor, 28 scraper, 29 ball, 30 partition plate, 31 cushion, 32 slide rail, 33 slide groove, 34 limit block, 35 buffer, 36 bearing plate. DETAILED DESCRIPTION

[0050] The technical solution of this patent is further described in detail below in conjunction with specific implementation methods.

[0051] The embodiments of the present invention are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0052] In the description of this patent, it should be understood that the terms "center", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the drawings, and are only for the convenience of describing this patent and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on this patent.

[0053] In the description of this patent, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", and "set" should be understood in a broad sense, for example, it can be fixedly connected or set, or it can be detachably connected or set, or connected or set in one piece. For ordinary technicians in this field, the specific meanings of the above terms in this patent can be understood according to specific circumstances.

[0054] Embodiment 1:

[0055] A facial expression recognition method based on intelligent care, such as Figure 1 As shown, the following steps are included:

[0056] S1: Collect static facial images of the person who needs care in advance;

[0057] S2: preprocessing the collected facial static images;

[0058] S3: The pre-processed facial image is divided into equal blocks and the texture feature points in the facial image are extracted using the LBP algorithm;

[0059] S4: According to the facial animation parameter rules obtained by statistical learning, the texture feature points are deformed with Bezier surface control to simulate and generate dynamic facial expression images;

[0060] S5: Packaging and transmitting the obtained single-category image data to the nursing information database for storage;

[0061] S6: monitor the person in care in real time, capture the face range based on the center of the two eyes and capture the face image;

[0062] S7: extracting LBP expression feature values ​​of the facial image through the LBP algorithm, comparing the extracted LBP expression feature values ​​with the parameter values ​​in the nursing information database, and obtaining the facial expression recognition result;

[0063] S8: Automatically determine whether the expression recognition result causes an alarm based on the warning mark in the care information database.

[0064] The method for collecting the facial static image includes but is not limited to picture entry, real-time photography, video capture, etc.

[0065] The preprocessing includes size normalization, noise reduction, grayscale processing and other methods.

[0066] Furthermore, the unit of the size normalization process is pixel.

[0067] Furthermore, the noise reduction process is one of linear mean filtering or Gaussian filtering.

[0068] The method for equally partitioned block processing comprises the following steps:

[0069] S101: Divide the detected facial image into small blocks of n×n pixel units;

[0070] S102: Calculate the LBP value in each small block, and connect the LBP values ​​in all small blocks into a feature vector to obtain the LBP texture feature vector of the detected facial image.

[0071] Further, the calculation of the LBP value in each small block includes the following steps:

[0072] S201: Compare the central pixel value of each small block with the 8 pixel values ​​of its neighborhood, and assign 1 to the pixel points that are ≥ the central pixel value, and assign 0 to the pixel points that are < the central pixel value;

[0073] S202: Binary encoding is performed clockwise starting from the upper left corner;

[0074] S203: Convert the encoded eight-bit binary number into a decimal number as the LBP value of the current pixel.

[0075] The generating of dynamic facial expression images specifically comprises the following steps:

[0076] S301: using the Bezier surface control deformation method and facial animation parameter rules, generating a golden skull mesh from texture feature points extracted from the facial image;

[0077] S302: Mapping the preprocessed facial image as a texture to the golden skull mesh, and controlling the expression change by using the image's gradient domain information and parameter mapping method;

[0078] S303: Using Open GL rendering, a facial image with various basic expressions is obtained.

[0079] Furthermore, the various basic expressions include safe expressions, warning expressions and dangerous expressions, and the safe expressions, warning expressions and dangerous expressions all include micro-expressions with different facial changes. According to the expression recognition results, medical staff are prompted to reduce the safety risks of the cared personnel.

[0080] Furthermore, the safe expressions include "joy" and "calmness", which are specifically manifested as micro-expressions such as upturned eyes, nose and mouth corners, and relaxed facial muscles;

[0081] The warning expressions include "sadness", "sadness" and "sleep status", etc. The specific manifestations of sadness are frowning, closing eyes in pain or tears, facial sweating, trembling lips, red face or eyes, flaring nostrils and other micro-expressions; the specific manifestations of sorrow are elongated face, especially loose and drooping facial muscles, frowning and tilted eyebrows and eyes and other micro-expressions; the specific manifestations of sleep status are closed eyes but frowned eyebrows, constantly closing mouth, sweating on the face, shaking head and other micro-expressions.

[0082] The dangerous expressions include "anger" and "shock", etc. The specific manifestations of anger are red or purple face, blue veins on the forehead, rapid breathing, open nostrils and trembling, tightly closed lips, clenched teeth, wide eyes, frowning and raised eyebrows, etc.; the specific manifestations of shock are pale face, raised eyebrows and open eyes and mouth, erect hair, trembling muscles on both sides, etc.

[0083] The single-category image data is static facial images and various dynamic facial expression images derived from static facial images.

[0084] The step of comparing the extracted LBP expression feature value with the parameter value in the nursing information database specifically includes the following steps:

[0085] S401: Compare the LBP expression feature value with the feature values ​​of various basic expressions, and confirm that the LBP value repetition point is greater than the set threshold, and the facial image of the caregiver with the most LBP value repetition points;

[0086] S402: Confirm the expression classification of the extracted LBP expression feature value according to the dynamic facial expression image. When in use, collect the facial image of the caregiver, extract the LBP expression feature value of the facial image, compare the extracted LBP expression feature value with the parameter value in the care information database, thereby locking the caregiver, and confirming the coverage information of the collected caregiver's facial image according to the dynamic facial expression image.

[0087] The nursing information database also includes a deletion module for deleting expired nursing staff image information, so as to clean up the occupied space in time.

[0088] The warning mark includes a warning level for distinguishing alarm information and an editing block for medical staff to input micro-expressions that each caregiver needs to be reminded of, so as to automatically confirm the alarm status of face recognition results of different caregivers according to the warning level and the editing block content.

[0089] Furthermore, the warning levels include one type of no alarm, two types of selective alarm and three types of direct alarm, which correspond to the safety expression, warning expression and danger expression respectively.

[0090] Before use, this embodiment collects a static facial image of the person being cared for by a static or dynamic method, and pre-processes the collected static facial image to improve clarity; the pre-processed facial image is processed into equal blocks, and the texture feature points in the facial image are extracted by using the LBP algorithm to ensure the accuracy of later face recognition; the texture feature points extracted from the facial image are generated into a golden skull grid by using the Bezier surface control deformation method and facial animation parameter rules, and the pre-processed facial image is mapped as a texture to the golden skull grid, and the expression change is controlled by using the image gradient domain information and parameter mapping method, and the curvature change of facial features such as the nose, cheeks, corners of the mouth, eyes, etc. is automatically rehearsed, and then Open GL is used for rendering to obtain a vivid image with various basic expressions; the obtained static facial image of each person being cared for and various dynamic facial expression images derived from the static facial image are packaged and transmitted to the care information database for storage;

[0091] During care, the LBP expression feature value is calculated based on the captured image, and the facial static image that fits the LBP expression feature value is searched in the care information database to confirm the identity information of the person being cared for. Based on various dynamic facial expression images derived from the facial static image, the expression classification of the extracted LBP expression feature value is confirmed, and the content of the patient's expression is analyzed to obtain the actual situation of the patient. If the monitored facial expression belongs to the third category of direct alarm level and meets the corresponding editing block content, an alarm is triggered to remind the medical staff to quickly reach the monitoring point; if the monitored facial expression belongs to the second category of selective alarm level, but does not meet the corresponding editing block content, then there is no need to issue an alarm. On the contrary, if the monitored facial expression belongs to the second category of selective alarm level and meets the corresponding editing block content, an alarm is triggered to remind the medical staff to quickly reach the monitoring point; if the monitored facial expression belongs to a safe expression but meets the corresponding editing block content, an alarm is triggered to remind the medical staff, effectively saving the medical staff's care energy and time, while ensuring the safety of the cared person.

[0092] Experiment: 20 to 50 people were selected, and 5 to 10 static facial images with various expressions were taken for each person, and then saved in the nursing information database. Three of them were randomly selected as targets, and then 37 people and three target people were randomly selected and introduced into the scanning room to make different expressions. After capturing the facial information of 40 people and extracting the facial features of the images, the three target people were locked, and the accuracy rate was 100%; and the alarm could be confirmed based on the expressions of the three target people, and the error rate was 0.

[0093] Embodiment 2:

[0094] A nursing device based on the facial expression recognition method described in Example 1 of smart nursing is used to facilitate people with physical disabilities to go to the toilet; Figure 1-7 As shown, the nursing mechanism includes a bed body 1 fixed to the top of four legs 9 and a monitoring part. The top outer wall of the bed body 1 is fixed with a lower bed board 8 by bolts, and an upper bed board 4 is hinged on one side edge of the lower bed board 8; the bottom inner wall of the bed body 1 is fixed with a top support member 2 by bolts, and the top support member 2 can be one of an electric push column, a jack, a cylinder, a hydraulic cylinder, etc. The extended end of the top support member 2 and the bottom outer wall of the upper bed board 4 are rotatably matched through a connecting part, and the connecting part is fixed to the extended end of the top support member 2 by bolts. The monitoring unit is composed of a camera head A3 fixed to one end of the bed body 1 by a support and a camera head B6 fixed to one end of the lower bed board 8 by a bracket. Both the camera head A3 and the camera head B6 are connected to the control module for communication. The camera heads A3 and B6 respectively capture the facial expressions of the caregiver in lying and sitting positions in real time from multiple angles so as to compare them with the parameter values ​​in the care information database to obtain the facial expression recognition results. An automatic door opening and closing part is arranged on the inner wall of one side of the bed board 8, and the automatic door opening and closing part comprises a fixing plate 5, a motor 12 and a transmission gear 21. The outer wall of the fixing plate 5 is slidably connected to the inner wall of the lower bed board 8, and a rack is fixed to the outer wall of one side of the fixing plate 5 by bolts; the bottom outer wall of the motor 12 is fixed to the inner wall of the bed body 1 near the top by bolts; the bottom end of the transmission gear 21 is connected to the output end of the motor 12 by a connecting shaft, and the outer wall of the transmission gear 21 is meshed with the outer wall of the rack; The interior of the bed body 1 is composed of a cleaning part and a collecting part. Before use, the medical staff will completely insert the cleaned waste storage box 11 into the bottom of the bed body 1. When in use, first start the top support 2 and then push up the upper bed board 4, so that it tilts with the connection with the lower bed board 8 as the base point, so as to flexibly control the patient's supine lying posture, and then start the motor 12 to drive the transmission gear 21 to rotate, and drive the fixed plate 5 to slide back and forth in the lower bed board 8 through the meshing of the rack, so that the discharge port is opened, so that the patient's waste when going to the toilet can enter the collection part for storage;

[0095] Preferably, the top outer wall of the upper bed board 4 and the top outer wall of the lower bed board 8 are both bonded with a protective pad 7 to ensure the comfort of the patient when lying down.

[0096] Furthermore, the collecting part includes a receiving box 13, a waste storage box 11, an inlet 15, a baffle plate 24 and a weight reminder structure, wherein one side outer wall of the receiving box 13 is fixed to one side outer wall of the fixing plate 5 through a connecting block; the top end of the waste storage box 11 is plugged into the bottom outer wall of the bed 1; the inlet 15 is opened on the bottom outer wall of the bed 1, and the inner wall on the opposite side of the inlet 15 is provided with a slide rail 32; the outer walls on both sides of the baffle plate 24 are slidably connected to the inner walls of the two slide rails 32 through sliders, and the top outer wall of the baffle plate 24 close to one edge is connected to the bottom outer wall of the receiving box 13 through a supporting block; the weight reminder structure is composed of a containing group and a detection group. When the fixed plate 5 moves forward and backward, the material receiving box 13 is driven to move synchronously through the connecting block. At the same time, the material receiving box 13 drives the blocking plate 24 to move accordingly through the supporting block, so as to synchronously complete the opening of the feed port and the closing of the inlet 15 through the blocking plate 24, so as to facilitate the collection of excrement and prevent the odor in the waste storage box 11 from being emitted.

[0097] Furthermore, the containing group includes a cushion 31, two slide grooves 33, two limit blocks 34, a buffer 35 and a supporting plate 36, the two slide grooves 33 are respectively opened on the inner walls of the opposite sides of the bed body 1; the outer walls of one side of the two limit blocks 34 are respectively slidably connected to the inner walls of the two slide grooves 33 through sliders; the bottom outer wall of the supporting plate 36 is fixed to the top of the two limit blocks 34 by bolts; the two ends of the buffer 35 are respectively welded to the bottom outer wall of the supporting plate 36 and the bottom inner wall of the bed body 1; the top outer wall and the bottom outer wall of the cushion 31 are respectively welded to the bottom outer wall of the supporting plate 36 and the bottom inner wall of the bed body 1. When the material receiving box 13 is passively moved forward and backward, the soft pad 31 is first pulled and squeezed by the supporting plate 36 to move down, so that the supporting plate 36 and the bottom inner wall of the bed body 1 are kept horizontal. When the material receiving box 13 stops moving, the supporting plate 36 is just in the hollow position of the material receiving box 13 and fits with the inner wall around it, and it is guided to rise in the sliding cooperation of the limit block 34 in the slide groove 33 under the support of the buffer 35; on the contrary, when the automatic opening and closing part of the box door is closed after use, the fixed plate 5, the material receiving box 13 and the blocking plate 24 move synchronously in the opposite direction. When the material receiving box 13 moves in the opposite direction, the soft pad 31 is pulled and squeezed by the supporting plate 36 to move down again, so that the supporting plate 36 and the bottom inner wall of the bed body 1 are kept horizontal. When the material receiving box 13 stops moving, the supporting plate 36 is completely out of the wrapping range of the material receiving box 13, and the material discharge port is adjusted to be closed and the inlet 15 is opened, so that all the excrement on the supporting plate 36 is transferred to the waste storage box 11.

[0098] Preferably, the receiving box 13 is in a "U"-shaped structure, which facilitates the excrement to fall to the bottom of the receiving box 13 while preventing the excrement from splashing and contacting the patient's skin. A scraper 28 is welded on the inner wall of one side of the receiving box 13 near the bottom, and two or more through grooves are provided on the bottom outer wall of the scraper 28. By using the multiple teeth formed by the through grooves, when the patient's single toilet experience is finished, when the receiving box 13 is passively displaced above the inlet 15, the multiple teeth at the bottom of the scraper 28 are in layer-by-layer contact with the top surface of the carrier plate 36, so as to automatically and efficiently scrape the dirt on the top surface into the waste storage box 11.

[0099] Furthermore, the detection group includes a partition plate 30 and three groups of vertically arranged and structurally identical pressure components, the bottom outer wall of the partition plate 30 is welded to the bottom inner wall of the receiving box 13; the pressure component is composed of an elastic member 16, an elastic cover 17, a T-bar 26, a collision sensor 27 and a sphere 29, one side outer wall of the collision sensor 27 is fixed to the one side inner wall of the receiving box 13 by bolts, the model of the collision sensor 27 is DFROBOT, and the signal output end of the collision sensor 27 is electrically connected to the control module; the sphere The outer wall of the ball 29 is embedded in the inner wall of one side of the material receiving box 13; one end of the T-shaped rod 26 is welded to the outer wall of the ball 29, and the T-shaped rod 26 and the partition plate 30 form a sliding fit; the elastic member 16 is wrapped around the circumferential outer wall of the T-shaped rod 26, and the two ends of the elastic member 16 are welded to the outer wall of the ball 29 and the inner wall of one side of the partition plate 30 in sequence; one end of the elastic cover 17 in a state of wrapping the ball 29 is welded to the inner wall of one side of the material receiving box 13; after the carrying plate 36 enters the wrapping range of the material receiving box 13, it is pushed up by the action of the buffer 35 When the load plate 36 moves downward, the ball 29 on the side is squeezed to shrink, thereby causing the elastic member 16 to shrink, and the T-shaped rod 26 slides through the partition plate 30, so that one end surface of the T-shaped rod 26 contacts the collision sensor 27 in front of the horizontal position, and the collision sensor 27 sends a sensing signal to the control module in time, so that the load plate 36 can be used to carry more excrement during use. Perform corresponding actions; when the supporting plate 36 indirectly touches the collision sensor 27 on the first group of pressure-bearing components and the second group of pressure-bearing components, the control module determines whether to close the automatic opening and closing part of the door within XX time according to the facial image collected by the camera B6; when the supporting plate 36 indirectly touches the collision sensor 27 on the third group of pressure-bearing components, the control module determines that the automatic opening and closing part of the door is still not controlled to be closed according to the facial image collected by the camera B6, and the control module initiates three types of direct alarms to prompt medical staff to arrive at the scene in time to reduce safety hazards.

[0100] Preferably, the sizes of the three spheres 29 are increased from top to bottom, so as to decrease in a step-by-step manner according to the change of the load pressure of the load-bearing plate 36 .

[0101] Further preferably, the inner walls of the four sides of the bed body 1 near the top are bonded with an adsorption layer 23, and the adsorption layer 23 is provided to facilitate the removal of the odor of excrement.

[0102] In order to facilitate cleaning the inner side of the material receiving box 13; Figure 2 and Figure 3 As shown, the cleaning part includes a water tank 18, a flushing head 14 and an inclined cleaning head 25, the bottom outer wall of the water tank 18 is clamped to the bottom inner wall of the bed 1; one end of the flushing head 14 is fixed to the top inner wall of the bed 1 by bolts, the input end of the flushing head 14 is connected to the outer wall of one side of the water tank 18 through a water pipe, and the switch control end of the flushing head 14 is electrically connected to the control module; one end of the cleaning head 25 is fixed to the inner wall of one side of the bed 1 near the top by bolts, the input end of the cleaning head 25 is connected to the outer wall of one side of the water tank 18 through a conduit, and the switch control end of the cleaning head 25 is electrically connected to the control module;

[0103] Preferably, a guide baffle 22 is welded to the inner wall of one side of the bed body 1 located above the cleaning head 25 to prevent excrement from contacting the cleaning head 25 below while not affecting the cleaning head 25 from flushing the patient's buttocks; a storage box 10 is plugged into the outer wall of one side of the bed body 1, which can be used to place paper towels and other items for the patient to take easily.

[0104] Further preferably, an observation window 19 and a water inlet 20 are fixed in sequence on the outer wall of one side of the bed body 1 close to the water tank 18 by bolts; cleaning liquid is stored in the water tank 18 through the water inlet 20, and when it is detected that the patient has finished using the toilet, the cleaning head 25 is controlled to start so as to rinse the patient's buttocks, and then the patient takes a paper towel from the storage box 10 to wipe it by himself; after the end, when the material receiving box 13 moves to above the inlet 15, that is, directly below the flushing head 14, the flushing head 14 is controlled to start so as to perform all-round high-speed flushing of the material receiving box 13 to ensure the cleanliness of its inner wall and improve the patient's later use experience. During the use of the equipment, medical staff can check the liquid level changes in the water tank 18 at any time through the observation window 19, so as to add cleaning liquid in time.

[0105] Before using this embodiment, the medical staff will completely insert the cleaned waste storage box 11 into the bottom of the bed body 1, and store the cleaning liquid in the water tank 18 through the water inlet 20; when in use, the camera A3 and the camera B6 will respectively capture the facial expressions of the care recipient in lying and sitting positions at multiple angles and in real time, so as to compare with the parameter values ​​in the care information database to obtain the facial expression recognition results.

[0106] When the patient needs to go to the toilet, he first starts the top support 2 and then pushes up the upper bed board 4, so that it tilts with the connection with the lower bed board 8 as the base point, so as to flexibly control the patient's supine posture, and then starts the motor 12 to drive the transmission gear 21 to rotate, and drives the fixed plate 5 to slide back and forth in the lower bed board 8 through the meshing of the rack, so that the discharge port is opened, and the fixed plate 5 drives the material receiving box 13 to move synchronously through the connecting block. At the same time, the material receiving box 13 drives the baffle plate 24 to move therewith through the support block, so as to complete the opening of the discharge port and the closing of the inlet 15 through the baffle plate 24 synchronously, so as to facilitate the collection of excrement while preventing the odor in the waste storage box 11 from being emitted.

[0107] When the receiving box 13 is passively moved forward and backward, the cushion 31 first pulls and squeezes the supporting plate 36 downward, so that the supporting plate 36 is kept level with the bottom inner wall of the bed body 1. When the receiving box 13 stops moving, the supporting plate 36 is just in the hollow position of the receiving box 13 and fits with the inner wall around it. Moreover, under the support of the buffer 35, the supporting plate 36 is guided to rise in the sliding cooperation in the slide groove 33 through the limit block 34, so that the supporting plate 36 enters the wrapping range of the receiving box 13 and is pushed up to the top of the ball 29 at the top under the action of the buffer 35. As more excrement is carried during use, it is pressed downward by the weight pressure, and the first group of pressure components, the second group of pressure components and the third group of pressure components arranged vertically are squeezed in turn during the downward movement. Specifically, when the supporting plate 36 moves downward, it squeezes the side The sphere 29 shrinks inward, thereby causing the elastic member 16 to contract, and the T-bar 26 slides through the partition plate 30, so that one end surface of the T-bar 26 contacts the collision sensor 27 directly in front of it in the horizontal direction, and the collision sensor 27 promptly sends a sensing signal to the control module so as to perform corresponding actions; when the supporting plate 36 indirectly touches the collision sensors 27 on the first group of pressure-bearing components and the second group of pressure-bearing components, the control module determines whether to close the automatic opening and closing part of the door within XX time according to the facial image collected by the camera B6; when the supporting plate 36 indirectly touches the collision sensor 27 on the third group of pressure-bearing components, the control module determines that the automatic opening and closing part of the door is still not controlled to be closed according to the facial image collected by the camera B6, and the control module initiates three types of direct alarms to prompt medical staff to arrive at the scene in time.

[0108] When the monitoring unit detects that the patient has finished using the toilet, the cleaning head 25 is controlled to start to wash the patient's buttocks. Then the patient takes a tissue from the storage box 10 to wipe the patient's buttocks. After staying for a period of time, the automatic opening and closing part of the control box door is closed, so that the fixed plate 5, the material receiving box 13 and the blocking plate 24 move synchronously in the opposite direction. When the material receiving box 13 moves in the opposite direction, that is, when it is passively displaced above the inlet 15, the cushion 31 is pulled and squeezed to move the supporting plate 36 downward again, so that the supporting plate 36 and the bottom inner wall of the bed body 1 are kept in the same plane. During this period, the scraper 28 is connected to the top of the supporting plate 36 by multiple notches at the bottom. The surface is contacted layer by layer, and then the dirt on the top surface of the supporting plate 36 is automatically and high-quality scraped off into the waste storage box 11. When the receiving box 13 stops moving, the supporting plate 36 is completely out of the wrapping range of the receiving box 13, and the discharge port is adjusted to be closed and the inlet 15 is opened. At this time, the flushing head 14 is controlled to start, so as to perform all-round high-speed flushing of the receiving box 13 to ensure the cleanliness of its inner wall and improve the patient's later use experience. During the use of the equipment, medical staff can check the liquid level changes in the water tank 18 at any time through the observation window 19, so as to add cleaning fluid in time.

[0109] References and sources in this invention:

[0110] Huang Feifei, Research on face recognition based on LBP, Master's thesis of Chongqing University, 2009.5 "Steps for extracting LBP feature vector";

[0111] Yu Jiajun, "Computer Simulation" CSCD, Peking University Core Journal, 2015, No. 10, pp. 271-276, "Automatic Generation of Dynamic Expressions for Portrait Cartoons";

[0112] Wang Ke, Feature tracking based on facial animation parameter extraction, Computer Engineering, No. 19, 2005.

[0113] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A facial expression recognition method based on intelligent care, characterized in that: The steps include: S1: Collect static facial images of the person who needs care in advance; S2: preprocessing the collected facial static images; S3: The pre-processed facial image is divided into equal blocks and the texture feature points in the facial image are extracted using the LBP algorithm; S4: According to the facial animation parameter rules obtained by statistical learning, the texture feature points are deformed with Bezier surface control to simulate and generate dynamic facial expression images; S5: Packaging and transmitting the obtained single-category image data to the nursing information database for storage; S6: Using the care institution to monitor the cared person in real time, capturing the face range based on the center of both eyes and capturing the face image; S7: extracting LBP expression feature values ​​of the facial image through the LBP algorithm, comparing the extracted LBP expression feature values ​​with the parameter values ​​in the nursing information database, and obtaining the facial expression recognition result; S8: Automatically determine whether the expression recognition result causes an alarm based on the warning mark in the care information database; The facial static image acquisition method includes picture input, real-time photography, and video capture; The preprocessing includes size normalization processing, noise reduction processing and grayscale processing methods, and the unit of size normalization processing is pixel; The noise reduction process is one of linear mean filtering or Gaussian filtering; The generating of dynamic facial expression images specifically comprises the following steps: S301: using the Bezier surface control deformation method and facial animation parameter rules, generating a golden skull mesh from texture feature points extracted from the facial image; S302: Mapping the preprocessed facial image as a texture to the golden skull mesh, and controlling the expression change by using the image's gradient domain information and parameter mapping method; S303: using Open GL rendering to obtain facial images with various basic expressions; The single type of image data is static facial images and various dynamic facial expression images derived from static facial images; the nursing information database also includes a deletion module for deleting expired nursing staff image information.

2. A facial expression recognition method based on intelligent care according to claim 1, characterized in that: The method for equally partitioned block processing comprises the following steps: S101: Divide the detected facial image into small blocks of n×n pixel units; S102: Calculate the LBP value in each small block, and connect the LBP values ​​in all small blocks into a feature vector to obtain the LBP texture feature vector of the detected facial image.

3. The method for facial expression recognition based on intelligent care according to claim 1, characterized in that: The various basic expressions include safe expressions, warning expressions and dangerous expressions. The safe expressions, warning expressions and dangerous expressions all include micro-expressions with different ranges of facial changes.

4. The method for facial expression recognition based on intelligent care according to claim 1, characterized in that: The step of comparing the extracted LBP expression feature value with the parameter value in the nursing information database specifically includes the following steps: S401: Compare the LBP expression feature value with the feature values ​​of various basic expressions, and confirm that the LBP value repetition point is greater than the set threshold, and the facial image of the caregiver with the most LBP value repetition points; S402: According to the dynamic facial expression image, confirm the expression classification of the extracted LBP expression feature value.

5. The method for facial expression recognition based on intelligent care according to claim 1, characterized in that: The warning mark includes a warning level for distinguishing alarm information and an editing block for medical staff to input micro-expressions that each caregiver needs to be reminded of; the warning levels include a first category of no alarm, a second category of selective alarm, and a third category of direct alarm, which correspond to safety expressions, warning expressions, and danger expressions, respectively.

6. A nursing device for implementing the method according to any one of claims 1 to 5, characterized in that: The nursing device comprises a nursing mechanism, which comprises a bed body (1) fixed to the top of two or more legs (9) and a monitoring unit, wherein the top outer wall of the bed body (1) is fixedly connected to a lower bed board (8), and an upper bed board (4) is arranged on one side edge of the lower bed board (8); the bottom inner wall of the bed body (1) is fixedly connected to a top support member (2), and the extended end of the top support member (2) and the bottom outer wall of the upper bed board (4) are rotatably matched through a connecting portion, and the connecting portion is composed of an arc-shaped plate fixedly connected to the extended end of the top support member (2) and a concave seat fixedly connected to the bottom outer wall of the upper bed board (4); the monitoring unit is composed of a shooting head A (3) fixed to one end of the bed body (1) through a pillar, and a camera head A (3) fixed to one end of the bed body (1) through a bracket. The invention relates to a machine for carrying out a cleaning operation of the present invention, wherein the machine comprises a camera head B (6) at one end of the lower bed board (8); the interior of the bed body (1) is composed of a cleaning part and a collecting part; an inner wall of one side of the lower bed board (8) is provided with an automatic door opening and closing part, and the automatic door opening and closing part comprises a fixed plate (5), a motor (12) and a transmission gear (21); the outer wall of the fixed plate (5) is slidably connected to the inner wall of the lower bed board (8), and a rack is fixedly connected to the outer wall of one side of the fixed plate (5); the outer wall of the bottom of the motor (12) is fixedly connected to the inner wall of the bed body (1) near the top; the bottom end of the transmission gear (21) is connected to the output end of the motor (12) through a connecting shaft, and the outer wall of the transmission gear (21) is meshed with the outer wall of the rack.

7. A nursing device according to claim 6, characterized in that: The cleaning section comprises a water tank (18), a flushing head (14) and an inclined cleaning head (25); the bottom outer wall of the water tank (18) is arranged on the bottom inner wall of the bed (1); one end of the flushing head (14) is fixedly connected to the top inner wall of the bed (1), and the input end of the flushing head (14) is connected to the outer wall of one side of the water tank (18) via a water pipe; one end of the cleaning head (25) is fixedly connected to the inner wall of one side of the bed (1) near the top, and the input end of the cleaning head (25) is connected to the outer wall of one side of the water tank (18) via a conduit.

8. A nursing device according to claim 6, characterized in that: The collecting part comprises a material receiving box (13), a waste storage box (11), an inlet (15), a baffle plate (24) and a weight reminder structure, wherein one side outer wall of the material receiving box (13) is fixed to one side outer wall of the fixing plate (5) via a connecting block; the top end of the waste storage box (11) is arranged on the bottom outer wall of the bed (1); the inlet (15) is arranged on the bottom outer wall of the bed (1), and the inner wall on the opposite side of the inlet (15) is provided with a slide rail (32); the outer walls on both sides of the baffle plate (24) are slidably connected to the inner walls of the two slide rails (32) via sliders, and the top outer wall of the baffle plate (24) close to one edge is connected to the bottom outer wall of the material receiving box (13) via a supporting block; and the weight reminder structure is composed of a containing group and a detection group.

Images