A sitting posture recognition method based on input data

Abstract

The present invention relates to the field of information processing technology, and specifically to a sitting posture recognition method based on input data, comprising: making a sitting posture recognition model under laboratory conditions, selecting one of the proportions of pressure sensor detection values ​​corresponding to sitting upright as a standard reference for sitting upright; when a user uses the method for the first time, the user completes at least one sitting upright and collects the detection values ​​of the pressure sensors; calculating the proportion of the pressure sensor detection values, comparing them with the standard reference for sitting upright, calculating the difference between the proportion of each pressure sensor detection value and the standard reference for sitting upright, and using the difference as a correction value; superimposing the correction value on each record of the sitting posture recognition model, and updating the proportion of pressure sensor detection values ​​corresponding to the sitting posture; periodically collecting the detection values ​​of the pressure sensors; calculating the proportion of the pressure sensor detection values, comparing them with the sitting posture recognition model, and obtaining the user's sitting posture obtained by the current period detection. The beneficial technical effects of the present invention include: being able to adapt to the body shape differences of different users and improving the accuracy of sitting posture recognition.

Description

Technical Field

[0001] The present invention relates to the field of information processing technology, and in particular to a sitting posture recognition method based on input data. Background Art

[0002] Most people today spend their days sitting in a chair. These long hours inevitably lead to prolonged sitting posture. Maintaining an unhealthy sitting posture for extended periods can lead to a series of lumbar spine problems. Chairs with posture recognition capabilities are currently available on the market. These utilize contact-type pressure sensors, which can only measure whether the sensor is pressed but not the force exerted. Consequently, existing posture recognition solutions simply determine the presence or absence of pressure points across the hips to determine the user's sitting posture. This approach has several drawbacks: First, it cannot detect interference from other objects. For example, if other objects are placed on the chair, it may misidentify the user's sitting posture. Second, it lacks compatibility with people of different body types. For example, the force area and pressure applied to a thicker or thinner person may differ. Finally, different sitting postures may cause the pressure on the seat cushion to vary, but the location of the pressure points remains unchanged, ultimately resulting in no change in the measurement results. Therefore, research on more accurate posture recognition technology is necessary.

[0003] For example, Chinese patent CN108814616A, published on November 16, 2018, discloses a sitting posture recognition method and a smart seat. The sitting posture recognition method is applied to a smart seat, which includes a backrest and a seat cushion. The smart seat is provided with a plurality of pressure sensors. The method includes: collecting the pressure applied by the user to the plurality of pressure sensors; locating the effective pressure sensor from the plurality of pressure sensors; determining the effective area according to the effective pressure sensor; obtaining the area of ​​the effective area and the pressure of the effective area, and identifying the user's sitting posture according to the change in the area of ​​the effective area and the change in the pressure of the effective area. In this way, the technical solution can recognize the user's sitting posture and remind the user to correct the sitting posture according to the user's sitting posture problem. However, the technical solution is difficult to adapt to the differences between different users. Summary of the Invention

[0004] The technical problem to be solved by this invention is the current lack of sitting posture recognition technology that can adapt to the body shape differences of different users. This invention proposes a sitting posture recognition method based on input data. This method can adapt to the body shape differences between users by modifying the sitting posture recognition model, thereby improving the accuracy of sitting posture recognition.

[0005] To solve the above technical problems, the present invention adopts the following technical solution: a sitting posture recognition method based on input data, which is used for sitting posture recognition of a seat equipped with a sitting posture array pressure sensor in the seat cushion, comprising:

[0006] Under laboratory conditions, a sitting posture recognition model is produced, wherein the sitting posture recognition model records the corresponding relationship between the proportion of pressure sensor detection values ​​and sitting posture, and a proportion of pressure sensor detection values ​​corresponding to upright sitting is selected as a standard reference for upright sitting;

[0007] When the user uses the device for the first time, he / she should sit upright at least once and collect the detection values ​​of all pressure sensors in the array in the upright sitting posture;

[0008] Calculate the proportion of pressure sensor detection values, compare with the positive standard reference, calculate the difference between the proportion of each pressure sensor detection value and the positive standard reference, and use the difference as the correction value;

[0009] Adding the correction value to each record of the sitting posture recognition model to update the proportion of the pressure sensor detection value corresponding to the sitting posture;

[0010] When the user is not using the system for the first time, the detection values ​​of all pressure sensors in the array are periodically collected;

[0011] The proportion of the pressure sensor detection value is calculated and compared with the sitting posture recognition model to obtain the user sitting posture obtained by the current cycle detection.

[0012] Preferably, the method for calculating the proportion of pressure sensor detection values ​​includes:

[0013] Calculate the sum of all pressure sensor detection values, denoted as S;

[0014] Calculate the ratio of each pressure sensor detection value to the total value S, which is the pressure sensor detection value ratio.

[0015] Preferably, the method for making a sitting posture recognition model includes:

[0016] Try various sitting postures under laboratory conditions, repeat each sitting posture multiple times, collect pressure sensor detection values ​​and calculate the average;

[0017] Categorize sitting postures into several preset sitting postures;

[0018] The mean value of the pressure sensor detection value corresponding to the sitting posture is associated with the sitting posture as sample data;

[0019] Calculate the proportion of pressure sensor detection values ​​of each pressure sensor in the sample data;

[0020] Regularize the pressure sensor detection value ratio according to a preset accuracy, and associate the regularized pressure sensor detection value ratio with the sitting posture as an item of the sitting posture recognition model;

[0021] When the user uses it, the detection values ​​of all pressure sensors in the array are collected and the proportion of the pressure sensor detection values ​​is calculated;

[0022] Find the entry in the sitting posture recognition model that is closest to the calculated pressure sensor detection value ratio. The sitting posture recorded in the entry is the recognized sitting posture.

[0023] Preferably, before calculating the proportion of the pressure sensor detection values, the pressure sensor detection values ​​are normalized, and the normalized pressure sensor detection values ​​are used to calculate the proportion of the pressure sensor detection values.

[0024] Preferably, try rocking sitting under laboratory conditions, wherein rocking sitting means rotating the upper body to the side with the maximum degree of tilt after sitting upright.

[0025] Read the detection values ​​of the pressure sensors in the array during the rocking process to obtain the maximum measurable value of each pressure sensor during the rocking process;

[0026] Perform the rocking test multiple times and calculate the average of the maximum measurable values ​​of each pressure sensor during the rocking test as the final maximum measurable value.

[0027] The ratio of the pressure sensor detection value to the maximum measurable value is used as the normalized value.

[0028] Preferably, when the user uses the device for the first time, after completing at least one sitting upright, he or she completes at least one rocking sitting, and records the maximum value of the pressure sensor detection value during the rocking sitting process, which is recorded as the measured maximum value;

[0029] Calculate the ratio of the maximum measurable value of each pressure sensor to the actual maximum value as the correction factor;

[0030] The proportion of the pressure sensor detection values ​​recorded in the sitting posture recognition model is multiplied by a correction coefficient to obtain an updated sitting posture recognition model.

[0031] Preferably, when a user uses the device for the first time, the method for completing sitting upright includes:

[0032] The user sits in the chair, tries to sit in a correct posture, and performs periodic tests until data that meets the standard appears;

[0033] The user is prompted to meet the standard, and then the user adjusts to the most comfortable sitting position, and the adjustment range does not exceed the preset threshold;

[0034] The user maintains an upright sitting position for a preset period of time, collects the pressure sensor detection value, and calculates the change in the proportion of the pressure sensor detection value;

[0035] The change amount is used as a preference correction amount, and the ratio of the pressure sensor detection values ​​recorded in the sitting posture recognition model is added to the preference correction amount to obtain an updated sitting posture recognition model.

[0036] The beneficial technical effects of the present invention include: being able to adapt to the body shape differences of different users and improving the accuracy of sitting posture recognition; being able to effectively identify the placement of foreign objects and avoid false alarms; the pressure value detected by the pressure sensor can approximately reflect the pressure distribution around the pressure sensor, which can support reducing the number of pressure sensors and achieve the purpose of controlling costs.

[0037] Other features and advantages of the present invention will be disclosed in detail in the following specific embodiments and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The present invention will be further described below with reference to the accompanying drawings:

[0039] Figure 1 2 is a flow chart of a sitting posture recognition method according to an embodiment of the present invention.

[0040] Figure 2 The figure is a flow chart of a method for making a sitting posture recognition model according to an embodiment of the present invention.

[0041] Figure 3 Schematic diagram of the process of the detection value normalization method according to an embodiment of the present invention.

[0042] Figure 4 The figure is a flow chart of a method for obtaining a correction coefficient according to an embodiment of the present invention.

[0043] Figure 5 Schematic diagram of the flow of a method for obtaining a preference correction value according to an embodiment of the present invention. DETAILED DESCRIPTION

[0044] The following is an explanation and description of the technical solutions of the embodiments of the present invention in conjunction with the drawings of the embodiments of the present invention. However, the following embodiments are only preferred embodiments of the present invention and are not exhaustive. Based on the embodiments in the implementation manner, other embodiments obtained by those skilled in the art without creative work are all within the scope of protection of the present invention.

[0045] In the following description, terms such as "inside", "outside", "up", "down", "left", "right", etc. that indicate directions or positional relationships are only used to facilitate the description of the embodiments and simplify the description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operate in a specific direction. Therefore, they should not be understood as limiting the present invention.

[0046] A sitting posture recognition method based on input data is used for the sitting posture recognition of a seat with a sitting posture array pressure sensor installed in the seat cushion. Figure 1 ,include:

[0047] Step A01) Under laboratory conditions, a sitting posture recognition model is produced. The sitting posture recognition model records the relationship between the proportion of pressure sensor detection values ​​and the sitting posture, and a pressure sensor detection value proportion corresponding to upright sitting is selected as a reference for the upright sitting standard.

[0048] Step A02) When the user uses the device for the first time, the user completes at least one sitting posture and collects the detection values ​​of all pressure sensors in the array in the sitting posture;

[0049] Step A03) calculating the proportion of pressure sensor detection values, comparing them with the standard reference, and calculating the difference between the proportion of each pressure sensor detection value and the standard reference, using the difference as a correction value;

[0050] Step A04) adding the correction value to each record of the sitting posture recognition model, and updating the proportion of the pressure sensor detection value corresponding to the sitting posture;

[0051] Step A05) When the user is not using the device for the first time, periodically collect the detection values ​​of all pressure sensors in the array;

[0052] Step A06) Calculate the proportion of pressure sensor detection values, compare with the sitting posture recognition model, and obtain the user sitting posture obtained by the current cycle detection.

[0053] Since each user has a different weight and body shape, when using it for the first time, you need to record the upright sitting posture once. Based on the data obtained from this upright sitting detection, the sitting posture recognition model is corrected to improve the accuracy of sitting posture recognition. There are many ways to set up the sitting posture array. This embodiment uses a V-shaped array as an example. The V-shaped array includes two rows of pressure sensors at an acute angle, and each row has 3 pressure sensors. The position of the V-shaped array corresponds to the position of the human thigh. According to the corresponding left thigh and right thigh, they are respectively recorded as the left column and the right column. The pressure sensors are numbered in sequence in the direction away from the human buttocks. That is, the left column is numbered as Left 1, Left 2 and Left 3, and the right column is numbered as Right 1, Right 2 and Right 3.

[0054] The method for calculating the pressure sensor detection value ratio includes: calculating the sum of all pressure sensor detection values, denoted as S; and calculating the ratio of each pressure sensor detection value to the sum S, which is the pressure sensor detection value ratio. Table 1 shows the pressure sensor detection value ratio calculated in this embodiment. Table 1 shows an upright sitting posture slightly forward, which is classified as an upright sitting posture.

[0055] Table 1 Proportions of pressure sensor detection values ​​calculated in this embodiment

[0056] serial number Detection value Proportion Regular proportion Left 1 400 0.108695652 0.10 Left 2 600 0.163043478 0.15 Left 3 900 0.244565217 0.25 Right 1 360 0.097826087 0.10 Right 2 580 0.157608696 0.15 Right 3 840 0.22826087 0.25 Sum 3680 1 1

[0057] Please see the attached Figure 2 ,The method of making a sitting posture recognition model includes:

[0058] Step B01) Trying multiple sitting postures under laboratory conditions, repeating each sitting posture multiple times, collecting pressure sensor detection values ​​and calculating the average;

[0059] Step B02) classifying the sitting posture into a plurality of preset sitting postures;

[0060] Step B03) associating the mean of the pressure sensor detection values ​​corresponding to the sitting posture with the sitting posture as sample data;

[0061] Step B04) calculating the proportion of pressure sensor detection values ​​of each pressure sensor in the sample data;

[0062] Step B05) normalizing the pressure sensor detection value ratio according to a preset accuracy, and associating the normalized pressure sensor detection value ratio with the sitting posture as an item of the sitting posture recognition model;

[0063] Step B06) When the user uses the device, the detection values ​​of all pressure sensors in the array are collected and the proportion of the pressure sensor detection values ​​is calculated;

[0064] Step B07) Find the entry in the sitting posture recognition model that is closest to the calculated pressure sensor detection value ratio. The sitting posture recorded in this entry is the recognized sitting posture. Preset sitting postures include sitting upright, left leg crossed, right leg crossed, sitting forward, sitting backward, and sitting vacant. The sitting upright posture can also be subdivided into various body shapes, such as sitting upright slightly leaning forward, sitting upright slightly leaning back, and sitting upright.

[0065] Before calculating the pressure sensor detection value ratio, the pressure sensor detection value is normalized and the pressure sensor detection value ratio is calculated using the normalized pressure sensor detection value. Figure 3 , the sitting posture recognition further includes: step C01) attempting to sit in a rocking position under laboratory conditions, where rocking means that after sitting upright, the upper body is tilted to the side and rotated in a circle with the maximum amplitude;

[0066] Step C02) reading the detection values ​​of the pressure sensors in the array during the rocking process to obtain the maximum measurable value of each pressure sensor during the rocking process;

[0067] Step C03) performing rocking multiple times, calculating the average of the maximum measurable values ​​of each pressure sensor during the rocking process as the final measurable maximum value;

[0068] Step C04) The ratio of the pressure sensor detection value to the maximum measurable value is used as a normalized value. The maximum pressure value that can be detected by each pressure sensor can be detected during rocking.

[0069] To further improve the accuracy of sitting posture recognition, please refer to the attached Figure 4, including: step D01) when the user uses the device for the first time, after completing at least one sitting upright, completes at least one rocking sitting, and records the maximum value of the pressure sensor detection value during the rocking sitting process, which is recorded as the measured maximum value;

[0070] Step D02) calculating the ratio of the maximum measurable value of each pressure sensor to the maximum measured value as a correction coefficient;

[0071] Step D03) Multiply the percentage of pressure sensor detection values ​​recorded in the sitting posture recognition model by a correction coefficient to obtain an updated sitting posture recognition model. Each pressure sensor is located in a different position, and the maximum value that can be detected is also different. Adjusting the correction coefficient can improve the accuracy of sitting posture recognition.

[0072] Please see the attached Figure 5 When the user uses it for the first time, the method to complete sitting upright includes:

[0073] Step E01) The user sits in a chair and attempts to sit in an upright position, and periodically checks until data that meets the standard appears;

[0074] Step E02) prompting the user to meet the standard, and then the user adjusts to the most comfortable upright position, with the adjustment range not exceeding a preset threshold;

[0075] Step E03) The user sits upright for a preset period of time, collects pressure sensor detection values, and calculates a change in the proportion of the pressure sensor detection values;

[0076] Step E04) The change is used as a preference correction. The pressure sensor detection value ratio recorded in the sitting posture recognition model is added to the preference correction to obtain the updated sitting posture recognition model. After the preference data is entered, the upright sitting standard reference is modified to meet the user's preferences.

[0077] The beneficial technical effects of this embodiment include: being able to adapt to the body shape differences of different users and improving the accuracy of sitting posture recognition; being able to effectively identify the placement of foreign objects and avoid false alarms; the pressure value detected by the pressure sensor can approximately reflect the pressure distribution around the pressure sensor, which can support reducing the number of pressure sensors and achieve the purpose of controlling costs.

[0078] The above are only specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Those skilled in the art should understand that the present invention includes but is not limited to the contents described in the drawings and the above specific embodiments. Any modifications that do not deviate from the functional and structural principles of the present invention are intended to be included within the scope of the claims.

Claims

1. A sitting posture recognition method based on input data, used for sitting posture recognition of a seat with a sitting posture array pressure sensor installed in the seat cushion, characterized in that: include: Under laboratory conditions, a sitting posture recognition model is produced, wherein the sitting posture recognition model records the corresponding relationship between the proportion of pressure sensor detection values ​​and sitting posture, and a proportion of pressure sensor detection values ​​corresponding to upright sitting is selected as a standard reference for upright sitting; When the user uses the device for the first time, he / she should sit upright at least once and collect the detection values ​​of all pressure sensors in the array in the upright sitting position; Calculate the proportion of pressure sensor detection values, compare with the positive standard reference, calculate the difference between the proportion of each pressure sensor detection value and the positive standard reference, and use the difference as the correction value; Adding the correction value to each record of the sitting posture recognition model to update the proportion of the pressure sensor detection value corresponding to the sitting posture; When the user is not using the system for the first time, the detection values ​​of all pressure sensors in the array are periodically collected; Calculate the proportion of pressure sensor detection values, compare with the sitting posture recognition model, and obtain the user sitting posture obtained by the current cycle detection; When a user uses it for the first time, the method to complete sitting upright includes: The user sits in the chair, tries to sit in a correct posture, and performs periodic tests until data that meets the standard appears; The user is prompted to meet the standard, and then the user adjusts to the most comfortable sitting position, and the adjustment range does not exceed the preset threshold; The user maintains an upright sitting position for a preset period of time, collects the pressure sensor detection value, and calculates the change in the proportion of the pressure sensor detection value; The change amount is used as a preference correction amount, and the ratio of the pressure sensor detection values ​​recorded in the sitting posture recognition model is added to the preference correction amount to obtain an updated sitting posture recognition model.

2. The method for sitting posture recognition based on input data according to claim 1, characterized in that: Methods for calculating the proportion of pressure sensor detection values ​​include: Calculate the sum of all pressure sensor detection values, denoted as S; Calculate the ratio of each pressure sensor detection value to the total value S, which is the pressure sensor detection value ratio.

3. The method for sitting posture recognition based on input data according to claim 2, characterized in that: Methods for making a sitting posture recognition model include: Try various sitting postures under laboratory conditions, repeat each sitting posture multiple times, collect pressure sensor detection values ​​and calculate the average; Categorize sitting postures into several preset sitting postures; The mean value of the pressure sensor detection value corresponding to the sitting posture is associated with the sitting posture as sample data; Calculate the proportion of pressure sensor detection values ​​of each pressure sensor in the sample data; Regularize the pressure sensor detection value ratio according to a preset accuracy, and associate the regularized pressure sensor detection value ratio with the sitting posture as an item of the sitting posture recognition model; When the user uses it, the detection values ​​of all pressure sensors in the array are collected and the proportion of the pressure sensor detection values ​​is calculated; Find the entry in the sitting posture recognition model that is closest to the calculated pressure sensor detection value ratio. The sitting posture recorded in the entry is the recognized sitting posture.

4. A sitting posture recognition method based on input data according to claim 2 or 3, characterized in that: Before calculating the pressure sensor detection value ratio, the pressure sensor detection value is normalized, and the normalized pressure sensor detection value is used to calculate the pressure sensor detection value ratio.

5. The method for sitting posture recognition based on input data according to claim 4, characterized in that: Try rocking in a sitting position under laboratory conditions. This involves sitting upright and then rotating the upper body sideways with the maximum possible range. Read the detection values ​​of the pressure sensors in the array during the rocking process to obtain the maximum measurable value of each pressure sensor during the rocking process; Perform the rocking test multiple times and calculate the average of the maximum measurable values ​​of each pressure sensor during the rocking test as the final maximum measurable value. The ratio of the pressure sensor detection value to the maximum measurable value is used as the normalized value.

6. The method for sitting posture recognition based on input data according to claim 5, characterized in that: When the user uses the device for the first time, they must complete at least one sitting upright and one rocking sitting position. The maximum value of the pressure sensor detected during the rocking sitting position is recorded as the measured maximum value. Calculate the ratio of the maximum measurable value of each pressure sensor to the actual maximum value as the correction factor; The proportion of the pressure sensor detection values ​​recorded in the sitting posture recognition model is multiplied by a correction coefficient to obtain an updated sitting posture recognition model.

Images