CN112001286A - Neck pillow height adjustment method and device for sleeping posture recognition processing based on pressure image - Google Patents

Abstract

The present invention provides a method and device for adjusting the height of a neck pillow for sleeping posture recognition based on a pressure image. The device includes pillow skin, pressure sensor, horizontal plate, side plate, electric push rod, fixed column, rotating shaft, control unit, down filling layer, power module and pressure sensor array. The power module provides sensor and controller power, and the pressure sensor array collects Human body pressure image, the pressure sensor receives the head pressure data, the control unit comprehensively analyzes the human body pressure image information and the head pressure data, and controls the electric push rod to rotate the side plate to change the height of the neck pillow. The above-mentioned device and method simplifies feature calculation, improves the recognition speed and accuracy of prone and supine lying, and achieves the purpose of recognizing the sleeping posture through pressure images and adjusting the height of the neck pillow to protect the human body.

Description

Neck pillow height adjustment method and device for sleeping posture recognition processing based on pressure image

technical field

The invention relates to a sleeping posture recognition technology, in particular to a method and a device for adjusting the height of a neck pillow for performing sleeping posture recognition processing based on a pressure image.

Background technique

More and more studies have shown that supine pillow height and side pillow height are positively correlated with the occurrence of cervical spondylosis. Improper pillow height will increase the burden on the spine and affect the quality of rest. In the products that adjust the pillow height according to the body posture to protect the cervical spine, the existing sleeping posture recognition technology mainly includes image recognition through the camera and pressure recognition on the pressure sensor. For identification through the images collected by the camera, it is difficult to ignore the influence of quilts and other items and the influence of weak night light in the actual situation, and the installation of the product is difficult and affects the beauty of the house. The method of pressure sensor recognition solves the problems of quilt occlusion and dark night, but the existing algorithms often have shortcomings such as many recognition modes but low accuracy and slow operation speed.

SUMMARY OF THE INVENTION

Therefore, in order to further improve the effect and efficiency, the present invention provides a simplified sleeping posture recognition method. In the sleeping posture recognition method, according to the inclination of lying on one's back and lying on different sides during sleep, by calculating the pressure density change of the pressure sensor, combined with the pressure distribution information and the optimal pillow height information of the human body in different postures, a new method based on Sleeping position recognition method based on pressure image processing, and obtained the relationship between the optimal pillow height and different sleeping positions.

A method for adjusting the height of a neck pillow for sleeping posture recognition processing based on a pressure image, comprising the following steps:

Step S1: determine the optimal pillow height and adjustment method;

Establish the optimal pillow height prediction function Y:

Among them, Y ₁ is the best pillow height when there is no cervical spondylosis, Y ₂ is the best pillow height when there is cervical spondylosis, Y ₃ is the pillow height when lying on the back, Y ₄ is the pillow height when lying on the side, and C is whether there is cervical spondylosis , C is 0 for no cervical spondylosis, C is 1 for cervical spondylosis, λ is the degree of adjustment of the height of the cervical pillow, and F is the adjustment function;

The user can select the adjustment mode function F by changing the adjustment time t _s and the power γ:

x is the time used to adjust to the current height during the height adjustment process, and t _s is the total time used for height adjustment, which is a constant;

Step S2: Preliminarily determine whether the human body is lying on its back and lying on its side;

S21. The pressure detection array composed of M*W pressure sensors collects the pressure information of the human body on the bed, and converts the obtained pressure value into a gray value of 0-255, and the conversion expression is:

Among them, N is the pressure value of the pressure sensor, N _max is the maximum pressure value, N _min is the minimum pressure value, and P is the gray value;

S22. Calculate the pressure density ρ:

where Ni is the pressure value, and P _ij is the _number of pressure points;

Double thresholds K ₁ and K ₂ are introduced. When the pressure density ρ of the pressure image is greater than K ₁ , the human body is lying on its back; if the pressure density ρ of the pressure image is less than K ₂ , the human body is lying on its side;

Step S3: Recognize that the human body is lying on its back and lying on its back;

The pressure-intensive area is located by the pressure-intensive area localization method based on local symmetry and grayscale statistical features, and the center-symmetric transformation is performed on the area with high pressure density, and then the transformed image is obtained by weighted averaging with the original image, and the standard deviation C ₁ is obtained:

In the formula, M _i is each statistical point in the detection window, each point is denoted as 1, and N _ij is the pressure value of the point;

Traverse the location information of each area to obtain the center location information, mark the situation with the largest standard deviation of the upper and lower two, obtain two positioning ranges, and compare the positioning ranges, namely:

In the formula, S _up and S _are respectively the detection window area closest to the head center Z and the farthest detection window area, and a is the ratio of the upper and lower parts to the upper and lower areas. When a is greater than K ₆ and S _{and the} When the distance d _under S is less than K ₇ , the human body is in a supine position, and the rest are in a prone position, wherein d is the distance between the _upper and _lower central positions of S, K ₆ is 0.6, and K ₇ is equal to the upper body length H2;

Step S4 : establishing a model for changing the height of the head and neck with posture, and adjusting the optimal pillow height according to the human posture detected in the above steps.

Preferably, in said S2:

When the pressure density ρ of the pressure image is between K ₁ and K ₂ , it is impossible to accurately distinguish whether the human body is lying on its back or on its side. Balance standard to judge the state of human body posture;

Set the thresholds K ₃ and K ₄ , and perform bottom hat transformation on the pressure image to remove the influence of noise. At the same time, select a number of pressure gray image location points whose gray value is greater than K ₅ , and select n*n points near the point to calculate the average difference , the average difference _C3 is

In the formula, P _p is the average value of the gray value of n*n points, P _ij is the gray value at the position (i, j), when C ₃ begins to be greater than the gradient threshold K ₃ , the point value is the contour point, the position of each contour point is counted, and the balance degree B of the pressure image at this time is:

B=|∑L _i -∑R _i | (8)

In the formula, Li is the contour point on the left side of the center line, and _Ri is the _contour point on the right side of the center line;

Construct the influence function r:

When r is from 0 to K ₃ , it is lying on the back, when r is from K ₄ to 0, it is lying on the side, and when r is at other values, the human body is not completely lying on the side.

Preferably, when the human body is in a supine position and the pressure density changes slowly within the time t, the height of the occiput is Y ₁ ; when the human body is in a supine position and the pressure density decreases rapidly, it is detected whether the human body is lying on the side, if within the time t, the occipital height is Y 1 ; The human body is still lying on the side and the pressure density changes slowly, adjust the pillow height to Y ₂ , if the human body is lying on the side and the pressure density changes slowly within the time t, adjust the pillow height to the lowest value Y ₃ ; when the human body is in the side lying position And the occipital height is Y ₂ when the pressure density changes slowly within the time t; when the human body is in a side lying position and the pressure density increases rapidly, it is detected whether the human body is lying on the back, if the human body is still lying on the back and the pressure density is still within the time t. If the change is slow, adjust the pillow height to Y ₁ , if the person is prone to lie down and the pressure density changes slowly during time t, adjust the pillow height to the lowest value Y ₃ .

A neck pillow height adjustment device for a neck pillow height adjustment method for sleeping posture recognition processing based on a pressure image, comprising a pillow skin, a pressure sensor, a horizontal plate, a side plate, an electric push rod, a fixed column, a rotating shaft, a control unit, a down Filling layer, power module, Bluetooth module and pressure sensor array, the electric push rod is connected to the fixed column and the side plate, the fixed column is connected to the horizontal plate, the control unit is connected to the electric push rod, the pressure sensor and the pressure sensor array, so The power module provides power to the sensor and the controller, the pressure sensor array collects human body pressure images for human sleep posture recognition, the pressure sensor receives head pressure data for head position recognition, and the control unit comprehensively analyzes the human body pressure images information and head pressure data, and by controlling the electric push rod to rotate the side plate to change the height of the neck pillow.

Preferably, the control unit determines the center position of the head according to the pressure of different pressure sensors, the device height l is negatively correlated with the pressure sensor pressure N, and the device height is positively correlated with the electric push rod length L, and the function formula is as follows:

l=αN+βL (12)

Among them, α and β are correlation coefficients, where α is positive and β is negative.

Compared with the prior art, the present invention has the following advantages:

(1) The method of identifying the sleeping posture by analyzing the pressure distribution of different postures of the human body solves the problem that the quilt occlusion and the dark light at night affect the sleeping posture recognition;

(2) In the range where the pressure density is relatively ambiguous for the recognition of sleeping posture, a constructor including the balance degree is introduced to accurately recognize it;

(3) The standard deviation judgment is introduced to simplify the feature calculation through the situation that the upper and lower pressure distribution ranges of the up and down lie are different;

(4) Finally, through the optimal pillow height model of sleeping posture and cervical spondylosis, the cervical vertebra protection through pressure image recognition of sleeping posture is realized.

Description of drawings

Fig. 1 is the pressure detection array diagram of the present invention;

Fig. 2a is the outline point schematic diagram of the balance degree of the present invention;

Fig. 2b is the value range schematic diagram of the constructor of the present invention;

3 is a schematic diagram of an apparatus according to an embodiment of the present invention;

Fig. 4 is the flow chart of the present invention;

FIG. 5 is an exemplary process diagram of the present invention performing sleep posture recognition; and

Figures 6a, 6b and 6c are schematic diagrams of three types of lying posture pressures of the present invention.

Reference number:

Pillow skin 10 , pressure sensor 20 , horizontal plate 21 , side plate 22 , electric push rod 23 , fixed column 24 , rotating shaft 25 , control unit 26 , down filling layer 27 , power module 28 , Bluetooth module 29 and pressure sensor array 30 .

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

It should be noted that, unless otherwise specified, the technical or scientific terms used in this application should have the usual meanings understood by those skilled in the art to which the present invention belongs.

In order to realize the idea of intelligently adjusting the support height of the head and neck, the present invention provides a cervical vertebra protection method and device for recognizing sleeping posture based on a pressure image. The process is shown in Figure 4.

Step S1: determine the optimal pillow height and adjustment method;

Specifically, in step S1, the information input by the user includes whether there is cervical spondylosis C, right palm width R, shoulder _- zygomatic distance S1, height H1, upper body length _H2 , shoulder width S2, neck pillow height adjustment degree λ, adjustment Mode F, lying pressure information N, etc. determine the optimal pillow height and adjustment mode. In the process of determining the optimal pillow height, according to the relationship between the right palm width, shoulder-zygomatic distance, height and weight and the optimal pillow height of the human body mentioned in the paper "Sleeping Pillow Height, Individual Data and Cervical Spondylosis" of Southeast University There is a certain relationship, and the corresponding correlation coefficients are 0.607, 0.603, 0.345, 0.378 to determine the pillow height model. In the application conclusion, the best pillow height for normal people without cervical spondylosis is the width of the right palm, which can be calculated according to the regression equation:

Y ₁ =1.558+0.807R (1)

The "same body size" data of the best pillow height for patients with cervical spondylosis is the shoulder-zygomatic distance, which can be calculated according to the regression equation:

Y ₂ =2.001+0.726S ₁ (2)

Among them, Y ₁ is the best pillow height when there is no cervical spondylosis, and Y ₂ is the best pillow height when there is cervical spondylosis. From this, individualized predictive values of optimal pillow height were obtained in patients with and without cervical spondylosis.

Further, according to whether the human body has cervical spondylosis or not, the λ value and the adjustment method function F suitable for the person are set. The optimal pillow height individualized prediction value is combined with the adjustment mode function to establish a function:

Among them, Y ₃ is the height of the pillow lying on the back, Y ₄ is the height of the pillow lying on the side, C is whether there is cervical spondylosis, C is 0 for no cervical spondylosis, C is 1 for cervical spondylosis, λ is the adjustment degree of cervical pillow height, F is the adjustment function.

Setting different adjustment methods function F can provide different adjustment methods. Optionally, set F as a power function, and users can choose their own comfortable adjustment method by changing the adjustment time t _s and the power γ:

x is the time used to adjust to the current height during the height adjustment process, and t _s is the total time used for height adjustment, which is a constant;

Step S2: Preliminarily determine the posture of the human body;

S21. As shown in FIG. 1, a pressure detection array composed of M*W pressure sensors collects the pressure information of the human body on the bed, and converts the pressure information into a pressure image and a grayscale image. Convert the obtained pressure value to a grayscale value of 0-255, the conversion expression is:

The specific method is:

Among them, N is the pressure value of the pressure sensor, N _max is the maximum pressure value, N _min is the minimum pressure value, and P is the gray value.

S22. Calculate the pressure density and introduce the double thresholds K ₁ and K ₂ . The calculation method of the pressure density ρ is:

Among them, Ni is the pressure value, and P _ij is the _number of pressure points.

Further, Fig. 6 is a schematic diagram of three kinds of lying posture pressure, Fig. 6a is lying on the side, Fig. 6b is lying on the back, Fig. 6c is lying on the prone, this figure shows the stress situation of the pressure sensor, wherein the darker the color, the greater the pressure, the human body. The force-bearing area of lying down or lying on your back is larger than the force-bearing area of lying on your side, but the mass of the human body remains unchanged, so the pressure density of lying on your stomach or lying on your back is smaller than that of lying on your side. When the pressure density ρ of the pressure image is greater than K ₁ , the human body is lying on its back; if the pressure density ρ of the pressure image is less than K ₂ , the human body is lying on the side; and when the pressure density ρ of the pressure image is between K ₁ Between K2 and K2, the human body is lying _on its back, not completely lying on its side, or already lying on its side. The values of K ₁ and K ₂ are determined by the supine pressure information S, and the S value is obtained by recording the pressure distribution information when the human body is lying flat, that is, the pressure density when the user is lying on a standard flat surface, and calculating the pressure density from this information. Optionally, K1 is less than _1.2 times S and K2 is between 1.5 times S and ₂ times S.

Further, in order to more accurately judge the state of the human body posture, the balance standard of the pressure image is introduced.

S23. Specifically, first determine the symmetry axis of the grayscale image according to the upper body length _H2 and the shoulder width S2. Taking the maximum force position of the pressure sensor 20 as the center Z, and the upper body length H2 as the distance, when there is pressure information at the position whose distance is H2 from Z, record the position and calculate the distance of each position, when the distance between the two points in all recorded positions is When the distance and shoulder width S ₂ are the closest, two points will be marked, and the line from the center of the two points to Z is the axis of symmetry.

Set the thresholds K ₃ and K ₄ , perform bottom hat transformation on the pressure image to remove certain noise effects, and then select a number of pressure gray image location points with gray values greater than K ₅ , and select n*n points near the point to average the difference Calculated, the average difference _C3 is

In the formula, P _p is the average value of the gray value of n*n points, P _ij is the gray value at the position (i, j), when C ₃ begins to be greater than the gradient threshold K ₃ , the point value is the contour point. The position of each contour point is counted, and the balance B of the pressure image is:

B=|∑L _i -∑R _i | (8)

In the formula, Li is the contour point on the left side of the center line, and _Ri is the _contour point on the right side of the center line.

The balance degree B contains the data information in three situations: lying on your back, not completely lying on your side, or lying on your side, as shown in the schematic diagram of the contour points of the balance degree in Figure 2a, which shows that the human body is not completely lying on its side and the force on both sides of the center line lower contour point. Further construct the influence function r:

As shown in Figure 2b, when r is from 0 to K ₃ , it is lying on its back, when r is from K ₄ to 0, it is lying on its side, and when r is at other values, the human body is not completely lying on its side, K ₃ and K _{4 The} value is related to B, and the value of _K5 is related to K1 and _K2 .

Further, through the above process, the posture of the human body can be preliminarily determined, that is, lying on one's back and lying on one's side.

Step S3: Recognize that the human body is lying on its back and lying on its back;

Specifically, in step S3, after the human body posture is preliminarily determined in step S2, the obtained reclining and reclining information is obtained by judging and analyzing, and the process is shown in the schematic diagram of sleeping posture recognition in FIG. 5 .

Specifically, in the process of judging the information about the body's upside-down and downsides, the pressure-intensive area is located by a pressure-intensive area location method based on local symmetry and grayscale statistical features. First of all, the gray value of the area with high pressure density changes greatly, so the standard deviation of this area is large. In the lying and lying pressure information, there are upper and lower regions with high pressure density, and this region has good upper and lower symmetry, so this region is transformed up and down symmetrically, and then the transformed image is obtained by weighted averaging with the original image. The standard deviation is calculated for the transformed image, and the area where the standard deviation is at the peak is the area with high pressure density to be identified. It should be noted that the mean value should be assigned to the position without pressure information before calculating the standard deviation. Specifically, referring to the head position Z, height H1, shoulder width S2 and the symmetry axis, the size is H1*S centered _on the symmetry axis. The detection area of ₂ , calculate the average value of this area and assign the average value to the position where the pressure is less than 10. The standard deviation _C1 is calculated as:

In the formula, M _i is each statistical point in the detection window, each point is denoted as 1, N _ij is the pressure value of the point, and the size of the statistical window is a square area that becomes the shoulder width S ₂ . The regions with the largest standard deviation were selected as candidate regions.

Traverse the position information of each area to obtain the center position information, and mark each case where the standard deviation of the peak is the largest. Combining the distances of each marked position to obtain two positioning ranges, and comparing the positioning ranges and distances respectively, the lying or lying information is obtained, namely:

In the formula, S- _up and S- _down are the detection window area closest to the head center Z and the farthest detection window area, respectively. As shown in Fig. 6c and Fig. 6b the pressure diagram of lying on the back, the upper box is S _-up , the lower box is S- _down , and a is the ratio of the area of the upper part to the sum of the area of the upper and lower parts. . When a is greater than K ₆ and the distance d between the _upper and _lower sides of S is smaller than K ₇ , the human body is in the supine position, and the rest are in the prone position. d is the distance between the _upper and _lower central positions of S, the value of K ₆ and K ₇ are related to the body weight distribution, here K ₆ is taken as 0.6, and K ₇ is taken as the value of the upper body length H2.

In the embodiment of recognizing the prone and supine postures of the human body, the number of pixels of the pressure image is equal to the number of pressure sensors, and the number of pixels is 500*280. Please refer to Fig. 6b, wherein Fig. 6b contains three standard deviations at the peaks, denoted as C ₁₁ , C ₁₂ and C ₁₃ , the sizes are 26.49, 33.37 and 37.11 respectively, the maximum distance is the distance of C ₁₁ and C ₁₃ , the distance d The size is 158. C ₁₁ is S _up , C ₁₃ is S _down , the range _on S is 24002, the range _below S is 12907, a is 0.65, and the upper body length H2 is 244, so the posture is lying on the back.

Figure 6c contains two standard deviations at the peaks, denoted as C ₂₁ and C ₂₂ , whose magnitudes are 33.98 and 35.06, respectively, the maximum distance is the distance between C ₂₁ and C ₂₂ , and the distance d is 274. C ₂₁ is S _up , C ₂₂ is S _down , the range _on S is 19317, the range _below S is 22172, a is 0.47, and the upper body length H2 is 244, so the posture is prone.

The positions _above S and _below S are shown in Fig. 6b and Fig. 6c.

S4: Establish a model of head and neck height changes with posture;

Specifically, in step S4, the optimal pillow height is adjusted according to the human body postures detected in the above steps, that is, lying on the side, lying on the prone, and lying on the back.

When the human body is in the supine position and the pressure density changes slowly within the time t, the height of the occiput is Y ₁ . When the human body is in the supine position and the pressure density decreases rapidly, it is detected whether the human body is lying on the side. When lying on the side and the pressure density changes slowly, adjust the pillow height to Y ₂ . If the human body is prone and the pressure density changes slowly within the time t, adjust the pillow height to the lowest value Y ₃ . When the human body is in the side lying position and the pressure density changes slowly within the time t, the height of the occiput is Y ₂ . When the human body is in the side lying position and the pressure density increases rapidly, it is detected whether the human body is lying on its back. In order to lie on the back and the pressure density changes slowly, adjust the pillow height to Y ₁ . If the person is prone to lie on the stomach and the pressure density changes slowly within the time t, adjust the pillow height to the lowest value Y ₃ .

Further, an embodiment of the device for adjusting the height of the head and neck is shown in FIG. 3, the structure includes a pillow skin 10, a pressure sensor 20, a horizontal plate 21, a side plate 22, an electric push rod 23, a fixed column 24, a rotating shaft 25, and a control unit 26. Down filling layer 27, power module 28, Bluetooth module 29, pressure sensor array 30, etc.

In the device for adjusting the height of the head and neck, the power module 28 is connected to the pressure sensor 20, the electric push rod 23, the control unit 26, the Bluetooth module 29 and the pressure sensor array 30, etc. The power module 28 provides power for the sensor and controller, and the pressure sensor array 30 collects human body pressure images, the pressure sensor 20 receives head pressure data, the Bluetooth module 29 transmits Bluetooth data to interact with terminals such as mobile phones, and the control unit 26 comprehensively analyzes the pressure sensor 26 information and pressure image information. The electric push rod 23 is connected to the fixed column 24 and the side plate 22 , and the fixed column 24 is connected to the horizontal plate 21 . The control unit 26 is connected to the electric push rod 23 , the pressure sensor 20 and the pressure sensor array 30 , and the pressure sensor array 30 receives the human body pressure information and transmits the information to the control unit 26 to recognize the sleeping posture of the human body. The pressure sensor 20 receives the head pressure data and transmits it to the control unit 26 for head position identification.

The control unit 26 causes the side plate 22 to rotate under the connection of the rotating shaft 25 by controlling the length of the electric push rod 23 . The rotation of the side plate 22 changes the internal force structure of the device and thus changes the height of the device under the force. When the height of the neck pillow needs to be increased, the electric push rod 23 near the center of the head is extended to cause the rotation shaft 25 to rotate and the side plate 22 to turn outwards, and the down filling layer 27 rises to raise the neck pillow. On the contrary, the neck pillow is lowered. The control unit 26 determines the center position of the head according to the pressure of different pressure sensors, the height of the device is negatively correlated with the pressure of the pressure sensor, and the height of the device is positively correlated with the length of the electric push rod 23 . Specifically, it can be as follows:

l=αN+βL (12)

Among them, l is the height of the device, N is the pressure value, L is the length of the electric push rod, α and β are the correlation coefficients, where α is positive and β is negative.

The above-mentioned embodiments are only used to illustrate the technical solutions of the present invention, but not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that it can still be used for the foregoing implementations. The technical solutions described in the examples are modified, or some or all of the technical features thereof are equivalently replaced; and these modifications or replacements do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present invention.

Claims (5)

1. a method for adjusting the height of a neck pillow based on a pressure image for sleeping posture recognition processing, is characterized in that, it comprises the following steps: Step S1: determine the optimal pillow height and adjustment method; Establish the optimal pillow height prediction function Y:

Among them, Y ₁ is the best pillow height when there is no cervical spondylosis, Y ₂ is the best pillow height when there is cervical spondylosis, Y ₃ is the pillow height when lying on the back, Y ₄ is the pillow height when lying on the side, and C is whether there is cervical spondylosis , C is 0 for no cervical spondylosis, C is 1 for cervical spondylosis, λ is the degree of adjustment of the height of the cervical pillow, and F is the adjustment function; The user can select the adjustment mode function F by changing the adjustment time t _s and the power γ:

x is the time used to adjust to the current height during the height adjustment process, and t _s is the total time used for height adjustment, which is a constant; Step S2: Preliminarily determine whether the human body is lying on its back and lying on its side; S21. The pressure detection array composed of M*W pressure sensors collects the pressure information of the human body on the bed, and converts the obtained pressure value into a gray value of 0-255, and the conversion expression is:

Among them, P is the gray value, N is the pressure sensor pressure, N _max is the maximum pressure, and N _min is the minimum pressure; S22. Calculate the pressure density ρ:

Among them, Ni is the pressure value, and P _ij is the _number of pressure points; Double thresholds K ₁ and K ₂ are introduced. When the pressure density ρ of the pressure image is greater than K ₁ , the human body is lying on its back; if the pressure density ρ of the pressure image is less than K ₂ , the human body is lying on its side; Step S3: Recognize that the human body is lying on its back and lying on its back; The pressure-intensive area is located by the pressure-intensive area localization method based on local symmetry and grayscale statistical features, and the center-symmetric transformation is performed on the area with high pressure density, and then the transformed image is obtained by weighted averaging with the original image, and the standard deviation C ₁ is obtained:

In the formula, M _i is each statistical point in the detection window, each point is denoted as 1, and N _ij is the pressure value of the point; Traverse the location information of each area to obtain the center location information, mark the situation with the largest standard deviation of the upper and lower two, obtain two positioning ranges, and compare the positioning ranges, namely:

In the formula, S _up and S _are respectively the detection window area closest to the head center Z and the farthest detection window area, and a is the ratio of the upper and lower parts to the upper and lower areas. When a is greater than K ₆ and S _{and the} When the distance d _under S is less than K ₇ , the human body is in a supine position, and the rest are in a prone position, wherein d is the distance between the _upper and _lower central positions of S, K ₆ is 0.6, and K ₇ is equal to the upper body length H2; Step S4 : establishing a model for changing the height of the head and neck with posture, and adjusting the optimal pillow height according to the human posture detected in the above steps. 2. The method for adjusting the height of a neck pillow for performing sleeping posture recognition processing based on a pressure image according to claim 1, wherein in the S2: When the pressure density ρ of the pressure image is between K ₁ and K ₂ , it is impossible to accurately distinguish whether the human body is lying on its back or on its side. Balance standard to judge the state of human body posture; Set the thresholds K ₃ and K ₄ , and perform bottom hat transformation on the pressure image to remove the influence of noise. At the same time, select a number of pressure gray image location points whose gray value is greater than K ₅ , and select n*n points near the point to calculate the average difference , the average difference _C3 is

In the formula, P _p is the average value of the gray value of n*n points, P _ij is the gray value at the position (i, j), when C ₃ starts to be larger than the gradient threshold K ₃ , the point value is the contour point, the position of each contour point is counted, and the balance degree B of the pressure image at this time is: B=|∑L _i -∑R _i | (8) In the formula, Li is the contour point on the left side of the center line, and _Ri is the _contour point on the right side of the center line; Construct the influence function r:

When r is from 0 to K ₃ , it is lying on the back, when r is from K ₄ to 0, it is lying on the side, and when r is at other values, the human body is not completely lying on the side. 3. The method for adjusting the height of a neck pillow according to claim 1, characterized in that: When the human body is in a supine position and the pressure density changes slowly within the time t, the height of the occiput is Y ₁ ; When the human body is lying on its back and the pressure density decreases rapidly, check whether the human body is lying on its side. If the human body is still lying on its side and the pressure density changes slowly within time t, adjust the pillow height to Y ₂ . In order to lie prone and the pressure density changes slowly, adjust the pillow height to the lowest value Y ₃ ; When the human body is in a side lying position and the pressure density changes slowly within the time t, the height of the occiput is Y ₂ ; When the human body is lying on the side and the pressure density increases rapidly, it is detected whether the human body is lying on its back. If the human body is still lying on its back and the pressure density changes slowly within the time t, adjust the pillow height to Y ₁ , if within the time t The human body is prone and the pressure density changes slowly, and the pillow height is adjusted to the lowest value Y ₃ . 4. A neck pillow height adjustment device for a neck pillow height adjustment method for sleeping posture recognition processing based on a pressure image, comprising a pillow skin, a pressure sensor, a horizontal plate, a side plate, an electric push rod, a fixed column, a rotating shaft, and a control unit , down filling layer, power module, bluetooth module and pressure sensor array, characterized in that: the electric push rod is connected to the fixed column and the side plate, the fixed column is connected to the horizontal plate, and the control unit is connected to the electric push rod, the pressure sensor and a pressure sensor array, the power module provides power to the sensor and the controller, the pressure sensor array collects human body pressure images for human sleep posture recognition, the pressure sensor receives head pressure data for head position recognition, and the control The unit comprehensively analyzes human body pressure image information and head pressure data, and controls the electric push rod to rotate the side plate to change the height of the neck pillow. 5. The device for adjusting the height of a neck pillow for sleeping posture recognition processing based on a pressure image according to claim 4, wherein the control unit determines the center position of the head according to the pressure of different pressure sensors, and the device height 1 is related to the pressure sensor. The pressure N is negatively correlated, and the device height is positively correlated with the electric push rod length L. The function formula is as follows: l=αN+βL (12) Among them, α and β are correlation coefficients, where α is positive and β is negative.

Images