A high-precision infrared temperature measurement method and device based on face recognition

Abstract

The present invention provides a high-precision infrared temperature measurement method based on face recognition, comprising the following steps: step 1, determining the size and area of ​​the forehead; step 2, determining the center point of the forehead and the allowable deviation; step 3, determining the distance triggering the temperature measurement according to the imaging; step 4, abnormal body temperature reminder judgment. The beneficial effect of the present invention is: combining the facial feature value to determine the moving position of the forehead center point coordinates on the screen, and then judging the distance by the imaging size, so that the actual coordinates and the effective temperature measurement coordinates meet a specific digital relationship, thereby accurately measuring the temperature.

Description

Technical Field

[0001] The invention belongs to infrared temperature measurement technology, and in particular relates to a high-precision infrared temperature measurement method based on face recognition. Background Art

[0002] Traditional non-contact infrared temperature measurement, such as forehead thermometers or infrared thermal imaging devices, can only collect thermal images or temperature values, and cannot associate the identity of the person being measured with the temperature value in real time. Manual registration is required, which may lead to cross-infection, data errors, and untimely reporting. At present, there are many infrared temperature measurement products based on face recognition technology on the market. They mainly measure temperature by adding infrared temperature measurement modules to face recognition devices, obtain the identity information of the person being measured through face recognition, and simultaneously associate the measured temperature and save it to the local or background system. However, such devices usually have the following problems:

[0003] Facial recognition devices mostly use infrared temperature sensors based on thermopile technology. Such sensors are mostly used in forehead and ear thermometers, so there are not many requirements for the distance, usually 1-3cm is enough. However, on facial recognition temperature measurement devices, if the distance is too close to measure the temperature, on the one hand, the user's face needs to be close to the screen, which is a very bad experience. On the other hand, if the distance is too close, the complete facial imaging will exceed the screen, and the facial feature values ​​cannot be effectively detected, resulting in facial recognition failure. Therefore, some facial devices will place the short-range temperature measurement sensor on the side of the device, use the arm to measure the temperature, or use a long metal gooseneck to protrude the sensor module from the device end, but both methods have the risk of poor experience and easy cross-infection.

[0004] The application of long-distance thermopile temperature sensors to facial devices can solve the problem of temperature measurement distance. However, due to the temperature measurement distance and accuracy requirements, facial temperature measurement mostly uses high-precision temperature sensors with a field of view angle of 3°-8°. At about 50cm away from the device screen, the effective temperature measurement diameter of the sensor is only about 2CM-4CM. In other words, a person must be able to accurately align his forehead with the designated temperature measurement point on the screen when he is 50cm away from the sensor, otherwise the collected temperature may deviate greatly. However, in actual situations, the person being measured does not know the exact distance between himself and the screen, and as the person gets closer and closer to the screen, the image of the person's forehead on the screen is constantly changing, resulting in the forehead temperature area for accurate temperature measurement not being accurately marked on the screen. If the mark is too small, it will make it difficult to accurately fit the forehead, resulting in accuracy errors. If the mark is too large, it will cause random fitting, and the sensor may collect the temperature of the hair or glasses, resulting in abnormal low temperature data. If the mark is too high, the temperature of the glasses or eyebrows may be collected when the forehead is actually fitted. If the mark is too low, the temperature of the hair may be collected. In short, there will be problems of temperature measurement error and unsatisfactory accuracy. Summary of the invention

[0005] The present invention studies the movement change trajectory of the human face on the temperature measurement screen during movement, calculates the center point and length and width of the forehead in combination with the face recognition feature value, determines the allowable effective temperature measurement area, and achieves the best measurement experience while ensuring the accuracy of temperature measurement, thereby improving the temperature measurement efficiency.

[0006] A high-precision infrared temperature measurement method based on face recognition, comprising:

[0007] Step 1: Determine the size and area of ​​the forehead;

[0008] Step 2: Determine the center point of the forehead and the allowable deviation;

[0009] Step 3: Determine the distance based on the imaging to trigger the temperature measurement;

[0010] Step 4: Abnormal body temperature alert judgment.

[0011] Furthermore, in step 1, a certain area on the forehead is selected as the forehead temperature zone, and the coordinates of the four points of the forehead temperature zone are recorded as: A(X1, Y1), B(X1, Y2), C(X2, Y1), D(X2, Y2), with the upper left corner of the display screen as the origin coordinate (0, 0), and the width D of the forehead temperature zone is recorded.

[0012] Furthermore, in step 2, the coordinate values ​​of the highest points of the two eyebrows on the face are obtained, which are T(Xt1, Yt1) and T(Xt2, Yt1), respectively. Then, (Xt1+Xt2) / 2 is used as the X coordinate of the forehead, and Yt1-D / 2 is used as the Y coordinate of the forehead. That is, when the person moves on the screen, the coordinates of the center point of the forehead are ((X1+X2) / 2, Y1-D / 2). Combined with the four identification coordinates of the effective temperature measurement imaging area of ​​the forehead model in step 1, it is necessary to satisfy Y2≥ Yt1≥Y1+D / 2 and X1≤(Xt1+Xt2) / 2≤X2.

[0013] Furthermore, in step 2, it also includes selecting 1 / 3 of the face frame as the forehead moving area to fit it with the effective forehead temperature mark, that is, comparing the relationship between the rectangular coordinates of the 1 / 3 area and the four coordinates of the forehead temperature mark on the screen to control the temperature measurement accuracy.

[0014] Furthermore, based on the camera imaging size, the imaging size at a certain distance is used as a trigger condition, face detection is enabled within the set distance, and a voice prompt is triggered to come closer.

[0015] Furthermore, when the forehead coordinates move within the forehead temperature zone, the sensor collects the temperature as the normal value, and then converts it into the body temperature value through the body surface temperature. If any of the forehead coordinates is not met, the temperature measurement is abnormal, and a voice prompt is activated to re-detect.

[0016] Furthermore, when the forehead coordinates move within the forehead temperature zone and Yt1<Y2 is satisfied, the sensor lens may collect the temperature of the person's hair and start a voice prompt for re-detection.

[0017] Furthermore, when the forehead coordinates move within the forehead temperature zone, if Yt1<Y1+D / 2, the temperature sensor may collect the temperature of the eyebrows or even the mask or glasses of the person being measured, resulting in a low temperature value, and a voice prompt is activated to re-detect.

[0018] A face recognition terminal that implements the above method adds an infrared temperature measurement module above the face recognition terminal and adds a forehead temperature zone mark on the device screen. When a person approaches the screen, the distance of the person from the screen is measured by the size of the person's face on the screen, and then the eyebrow center coordinates are calculated through the facial feature value coordinates of the two eyebrows.

[0019] The beneficial effect of the present invention is: combining the facial feature values ​​to determine the moving position of the forehead center point coordinates on the screen, and then determining the distance by the imaging size, so that the actual coordinates and the effective temperature measurement coordinates satisfy a specific digital relationship, thereby achieving an accurate temperature measurement method. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a structural schematic diagram of an infrared temperature measuring device provided in an embodiment of the present invention.

[0021] Figure 2 Schematic diagram of the effective temperature measurement imaging area in an embodiment of the present invention.

[0022] Figure 3 It is a schematic diagram of the explosion structure of the infrared temperature measuring device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0023] The embodiments of the present invention are described in detail below with reference to the accompanying drawings, but the present invention can be implemented in many different ways as defined and covered by the claims.

[0024] A high-precision infrared temperature measurement method based on face recognition based on the existing infrared temperature measurement terminal includes the following steps:

[0025] Step 1. Determine the size and area of ​​the forehead.

[0026] The width and length of foreheads of different people are different. In order to ensure that the temperature measurement algorithm is applicable to the vast majority of people, we actually measured people with different face sizes and took 4cm*8cm as a minimum forehead area, where 4cm is the width of the forehead, that is, the distance from the eyebrow line to the hairline, and 8cm is the length, that is, the distance between the two sides of the temple. Within this size range, the temperature values ​​collected by the sensor can be used as normal body temperature values.

[0027] We move the forehead model of 4cm*8cm in size at different distances in front of the infrared sensor and observe the actual image size of the forehead model on the screen. Since we need to take into account both the face detection and recognition distance and the effective temperature measurement distance of the forehead, we first filter out the inappropriate distance. According to the actual selected camera lens parameters, when the distance is 20cm from the screen, the facial image has completely covered or even exceeded the screen. This distance is not conducive to the extraction of facial feature values. When the distance is more than 50cm from the screen, it exceeds the distance for accurate temperature measurement by the sensor. Therefore, we compromised and selected 30-50cm as the effective temperature measurement distance. According to the sensor lens height and field of view, the effective temperature measurement area at 30cm, 40cm, and 50cm is calculated respectively, and then the corresponding forehead model screen image size corresponding to these three distances is observed. As people walk towards the screen from a distance, at 50cm, 40cm, and 30cm, the forehead model image will become larger and larger, and the position on the screen will be closer and closer to the top of the screen:

[0028] According to actual observations, the overlapped images of accurate temperature measurement at 30cm and 50cm foreheads are small. To ensure the consistency and accuracy of temperature measurement, we selected the 40-50cm overlapped area as the accurate screen temperature measurement mark and recorded the coordinates of the four points in the forehead temperature zone: A (X1, Y1), B (X1, Y2), C (X2, Y1), D (X2, Y2), with the upper left corner of the display screen as the origin coordinate (0,0). The width D of the forehead temperature zone is also recorded. See Figure 2.

[0029] Step 2: Determine the center point of the forehead and the allowable deviation.

[0030] We use the face detection algorithm to obtain the coordinate values ​​of the highest points of the two eyebrows on the face, which are T (Xt1, Yt1) and T (Xt2, Yt1), and then use (Xt1+Xt2) / 2 as the X coordinate of the forehead and Yt1-D / 2 as the Y coordinate of the forehead. That is, when a person moves on the screen, the coordinates of the center point of the forehead are ((X1+X2) / 2, Y1-D / 2), combined with the four identification coordinates of the effective temperature measurement imaging area of ​​the forehead model in step 1. It is necessary to satisfy Y2≥Yt1≥Y1+D / 2 and X1≤(Xt1+Xt2) / 2≤X2.

[0031] Step 3: Determine the distance based on the imaging to trigger the temperature measurement.

[0032] According to the camera imaging size, the imaging size at 50cm is used as the trigger condition. Face detection is turned on within 1 meter, and a voice prompt is triggered to move closer. When the person being measured approaches 50cm, the effective temperature measurement value is collected, and the face recognition algorithm is compared at the same time. When the person being measured is less than 40cm away from the screen, the voice prompt asks you to move back, that is, the effective temperature measurement of the person being measured has an adjustment distance of 10cm to avoid strict control of the distance caused by single-point temperature measurement. At the same time, because there are many overlapping areas in the imaging of the effective forehead temperature zone of 40-50, the prompt within this distance range is aimed at the forehead temperature zone, which can ensure that the infrared temperature measurement lens is completely aimed at the forehead, and the temperature measurement accuracy is the highest.

[0033] The area and size of the temperature measurement mark involved are related to the sensor selection. The 40-50cm temperature measurement is not unchangeable. The value of the face model selection size and the specific imaging size are all implemented on the specific face device implemented in the patent. It is related to the sensor and camera lens selected for the device. If you use sensors with other distances and field of view angles, or change the screen size and camera parameters, the overall logic of accuracy judgment involved in the patent is the same.

[0034] Step 4: Abnormal body temperature alert judgment.

[0035] When the forehead coordinates move within the range described in step 2, the sensor collects the temperature as the normal value, and then converts it into the body temperature value through the body surface temperature. If any of the forehead coordinates are not satisfied, the temperature measurement is abnormal. At this time, no matter how much the sensor takes, it will start the voice prompt to re-detect. Generally speaking, the allowed temperature measurement range on the left and right sides of the forehead is larger than that on the top and bottom, because the width of the forehead is limited. Yt1<Y2, which means that the forehead is generally lower, then the sensor lens may collect the temperature of the person's hair. If Yt1<Y1+D / 2, which means that the forehead is generally higher, then the temperature sensor may collect the temperature of the eyebrows or even the mask or glasses of the person being measured, which may lead to a low temperature measurement value.

[0036] In step 2, a relatively wide rectangular face frame is used, and the upper 1 / 3 part of the face frame is taken as the forehead moving area, so as to fit it with the effective forehead temperature mark, that is, the relationship between the coordinates of the 1 / 3 area rectangle and the four coordinates of the screen forehead temperature mark is compared to control the temperature measurement accuracy. This method is similar to the eyebrow center value described in the present invention. Only the size of the forehead value may be different;

[0037] According to another embodiment of the present method: a temperature measurement mark (such as a yellow foot) is pasted on the ground at a fixed position in front of the temperature measuring device. When the person being measured stands at this position, the distance between the person being measured and the screen is fixed, and then the effective forehead temperature area is marked on the screen. If the distance is constant, the marked effective temperature measurement area is fixed. At this time, as long as the forehead is in contact, the temperature measurement data is relatively accurate. This method controls the distance range that the person being measured moves in front of the screen through the landmark, and there is no need to judge the image size and distance, but its main temperature measurement logic is included in this patent.

[0038] According to the third embodiment of the present method: the forehead temperature measurement area is not limited to a square, and it is only determined whether the line connecting the two sides of the forehead is covered by the effective temperature measurement area on the screen. If covered, the temperature can be measured normally. This method is essentially to determine the relationship between the X and Y of the forehead coordinates and the identification coordinates.

[0039] like Figure 1-2 As shown, the face temperature measurement device adopted by the present invention is to add an infrared temperature measurement module 10 above the face recognition terminal, and add a forehead temperature zone mark 30 on the device screen. When a person approaches the screen, the binocular camera 20 obtains the facial graphics of the person, and measures the distance of the person from the screen by imaging the size of the person's face on the screen, and then calculates the eyebrow center coordinates through the facial feature value coordinates of the two eyebrows. The average length and width of the normal forehead is used as a reference model to calculate the imaging length and width of the forehead area when it moves on the screen at 30-50cm, and determine the overlapping area compatible with multiple distances, so as to determine the accurate position and size of the forehead temperature zone on the screen, and then compare it with the imaging coordinates of the center point of the forehead on the screen when the person moves, so as to accurately judge whether the forehead area is aligned with the sensor lens and give a voice prompt, so as to ensure the accuracy of the collected temperature.

[0040] like Figure 3 As shown, an infrared temperature measuring device of the present invention includes a housing 8, a display screen 1 is fixed on the housing 8, a mounting bracket 9 is fixed on the back of the housing, and a light shield 2 is fixed on the mounting bracket 9 and covers the upper end of the device. An infrared temperature measuring module 10 is fixed on the light shield 2, and the device also includes a binocular camera 20, a fill light 5, a main board 4, an interface circuit board 6, and a battery 3, which are installed in the housing. The binocular camera 20 is fixed to the upper end position in the housing 8 through a camera fixing plate 21, and the fill light 5 is fixed to the upper end position in the housing 8 through a fixing bar 11. The battery and the circuit board are fixed in the housing, and the interface circuit board 6 is connected to the main board circuit, fixed in the housing, and extends through the through hole behind the housing to be connected to an external device through an interface.

[0041] Screw holes 19 for fixing the light shield 2 are protruded from both sides of the fixing frame 9. The light shield 2 is fixed to the fixing frame by screws. The fixing frame is also installed on the back of the device.

[0042] like Figure 3 As shown, the device further includes a sensing module 12 and a speaker 13, which are arranged in the lower part of the device housing. The sensing module 12 is mainly arranged in a sensing circuit of a sensing circuit board, such as an RF radio frequency sensing circuit. The speaker 13 is fixed in the housing through a speaker bracket 23 and is connected to the audio circuit board 14. A card swiping area indicator mark 22 is marked on the display screen, and the IC card contacts the display screen at the mark to complete the card swiping.

[0043] The sensing module 12 and the speaker 13 are connected to the circuit board 4 through a circuit. The circuit board 4 and the sensing module 12 are both provided with heat sinks (16, 15) for dissipating heat from the chip on the circuit board.

[0044] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A high-precision infrared temperature measurement method based on face recognition, characterized in that: The following steps are involved: Step 1: Determine the size and area of ​​the forehead; Step 2: Determine the center point of the forehead and the allowable deviation; Step 3: Determine the distance based on the imaging to trigger the temperature measurement; Step 4: Abnormal body temperature reminder judgment; In step 1, a certain area on the forehead is selected as the forehead temperature zone, and the coordinates of the four points of the forehead temperature zone are recorded as A (X1, Y1), B (X1, Y2), C (X2, Y1), and D (X2, Y2). The coordinates are based on the upper left corner of the display screen as the origin (0, 0), and the width D of the forehead temperature zone is recorded; In step 2, the coordinate values ​​of the highest points of the two eyebrows on the face are obtained, which are T(Xt1, Yt1) and T(Xt2, Yt1), respectively. Then (Xt1+Xt2) / 2 is used as the X coordinate of the forehead, and Yt1-D / 2 is used as the Y coordinate of the forehead. That is, when a person moves on the screen, the coordinates of the center point of the forehead are ((X1+X2) / 2, Y1-D / 2). Combined with the four identification coordinates of the effective temperature measurement imaging area of ​​the forehead model in step 1, it is necessary to satisfy Y2≥Yt1≥Y1+D / 2 and X1≤(Xt1+Xt2) / 2≤X2.

2. According to the high-precision infrared temperature measurement method based on face recognition according to claim 1, it is characterized in that: In step 2, it also includes selecting 1 / 3 of the upper face frame as the forehead moving area, so as to fit it with the effective forehead temperature mark, that is, comparing the relationship between the rectangular coordinates of the 1 / 3 area and the four coordinates of the forehead temperature mark on the screen to control the temperature measurement accuracy.

3. According to the high-precision infrared temperature measurement method based on face recognition according to claim 1, it is characterized in that: Based on the camera imaging size, the imaging size at a certain distance is used as a trigger condition. Face detection is enabled within the set distance, and a voice prompt is triggered to ask the user to come closer.

4. According to the high-precision infrared temperature measurement method based on face recognition according to claim 1, it is characterized in that: When the forehead coordinates move within the forehead temperature zone, the sensor collects the temperature as the normal value, and then converts it into the body temperature value through the body surface temperature. If any of the forehead coordinates are not met, the temperature measurement is abnormal and a voice prompt is activated to re-detect.

5. A high-precision infrared temperature measurement method based on face recognition according to claim 4, characterized in that: When the forehead coordinates move within the forehead temperature zone and Yt1<Y2 is satisfied, the sensor lens may collect the temperature of the person's hair and start the voice prompt for re-detection.

6. The high-precision infrared temperature measurement method based on face recognition according to claim 4 is characterized in that: When the forehead coordinates move within the forehead temperature zone, if Yt1<Y1+D / 2, the temperature sensor may collect the temperature of the eyebrows or even the mask or glasses of the person being measured, resulting in a low temperature value, and a voice prompt is activated to re-detect.

7. The high-precision infrared temperature measurement method based on face recognition according to claim 1 is characterized in that: A face recognition terminal for implementing the method adds an infrared temperature measurement module above the face recognition terminal and adds a forehead temperature zone mark on the device screen. When a person approaches the screen, the distance of the person from the screen is measured by the size of the image of the person's face on the screen, and then the eyebrow center coordinates are calculated through the facial feature value coordinates of the two eyebrows.

Images