KR20220064629A - Temperature calibration method for measuring body temperature - Google Patents

Abstract

A temperature information calibration method for measuring a body temperature of the present invention includes: acquiring a fpa value map of a thermal image by using the thermal image and a fpa value of at least three pixels of the thermal image; associating fpa values of a general image of a specific point where a temperature value is disclosed and the thermal image corresponding to the specific point; and acquiring temperature distribution information or temperature information of the thermal image by using the temperature value of the specific point and the fpa value map.

Description

Temperature calibration method for measuring body temperature

The present invention relates to a method for correcting temperature information. Specifically, as a method of measuring body temperature, it relates to a method of measuring temperature without contact with a human body and correcting the temperature with more accurate temperature information. In the non-contact method, the temperature can be predicted by measuring the radiant energy of the human body.

One way to measure body temperature is to measure radiant energy. A thermal imaging camera may be used as a device for measuring the radiant energy.

The thermal imaging camera may estimate the temperature from the measured fpa value. However, the external temperature and humidity, the distance to the object, and the internal temperature of the camera act as variables. If the above variables are not taken into account, accurate body temperature measurement is impossible.

As a method of measuring a temperature using the thermal imaging camera, Citation 1 is known. Reference 1 uses the fpa value (0-16383) and performs a process of shifting the measurement value of the thermal imaging camera using the ambient temperature for accurate temperature measurement. Even according to the cited literature, there is a problem in that it cannot properly respond to the distance to the object, the internal temperature of the camera, and the external temperature.

Meanwhile, the thermal imaging camera provides the fpa value. The fpa value is not provided for all pixels. Of course, expensive devices provide all the fpa values, but many thermal imaging cameras do not provide numerical information of the fpa values at a level corresponding to pixels. In this case, it is difficult to obtain temperature information from the entire image, and especially to measure body temperature.

KR101731287B1: Method and apparatus for compensating output of infrared detector

The present invention is proposed against the background described above, and proposes a temperature information correction method for body temperature measurement that can obtain accurate body temperature information when information from a thermal imaging camera is limited.

The present invention proposes a temperature information correction method for body temperature measurement that provides accurate body temperature information by reflecting all external temperature and humidity, distance to an object, and internal temperature of a camera.

The method for correcting temperature information for body temperature measurement of the present invention comprises: obtaining an fpa value map of the thermal image using a thermal image and fpa values of at least three pixels of the thermal image; associating a general image of a specific point whose temperature value is known and an fpa value of the thermal image corresponding to the specific point; and obtaining temperature distribution information or body temperature information of a thermal image by using the temperature value of the specific point and the fpa value map.

According to another aspect, at least one of the fpa values may be a value of a center of the thermal image.

According to another aspect, at least one of the fpa values may be a highest value in the thermal image, and at least one of the fpa values may be a lowest value in the thermal image.

According to another aspect, the specific point may be a high temperature in the thermal image.

According to another aspect,

The fpa value map is obtained by Equations 1 and 2,

Equation 1:

Equation 2:

Here, fpa means fpa value, i and j mean the x and y-axis index of the pixel of the thermal image, gray means the luminance of the black-and-white image that is the thermal image, T means the luminance, and M is means max, m means min, and c means center.

According to another aspect, in order to correct an error according to a distance at which the thermal image is acquired, correcting the highest fpa value in the thermal image may be further performed.

According to another aspect, the highest fpa value may be an fpa value of a pixel having the highest temperature among the temperature distribution information.

According to another aspect, the correction may be performed by adding a certain correction value.

According to another aspect, the correction value may be obtained by using that the size of a box extracted by an extraction algorithm for extracting a part of the image becomes smaller according to the distance at which the thermal image is acquired.

According to the present invention, accurate body temperature information can be obtained irrespective of the external temperature and humidity, the distance to the object, and the internal temperature of the camera.

1 is a flowchart illustrating a temperature correction method according to a first embodiment.
FIG. 2 is a view schematically showing the process of FIG. 1 .
3 is a flowchart illustrating a temperature correction method according to the second embodiment.
FIG. 4 is a diagram showing a rough correlation between distance information between a camera and a body, the size of the box image, and an fpaM value.

Hereinafter, specific embodiments of the present invention will be described in detail with reference to the drawings. However, the spirit of the present invention is not limited to the following embodiments, and those skilled in the art who understand the spirit of the present invention can easily change other embodiments included in the scope of the same idea by adding, changing, deleting, and adding components. It will be possible to suggest, but it will be said that the embodiments are also included in the spirit of the present invention.

The temperature information correction method for body temperature measurement of the present invention may be implemented by a predetermined hardware device. The hardware may include a photographing device and a digital information processing device that processes information of the photographing device. The photographing apparatus may include a thermal imaging camera for photographing a thermal image, and a general camera for photographing a general color or black-and-white image. The digital information processing apparatus may include a processor for processing the thermal image and the general image, a memory in which necessary information is stored, and an IO device responsible for input/output of information. The temperature compensation method of the present invention may provide an image photographed at another location by simply processing it. It is also possible to quickly provide temperature information by taking an image in the field.

In an embodiment of the present invention, the thermal imaging camera may provide a thermal image. The thermal imaging camera can provide a limited number of fpa values. The thermal imaging camera may provide the highest fpa value, the lowest fpa value, and the image center fpa value. The thermal image may be a black-and-white image. The general camera may provide a general image in color.

In each of the following figures, fpa means fpa value, i and j mean the x and y-axis index of a pixel, gray means the luminance of a black-and-white image that is a thermal image, T means the luminance, and M means max, m means min, c means center, body means body, temp means temperature, box means box image extracted by a predetermined extraction algorithm, N is a general image, d is a distance, δ is a correction value, btd is a distance difference as much as d, dtf is a difference in fpa, and func is a function. .

The fpa value may have a range of 0-16383, and the luminance value may have a range of 0-256. The fpa value may have a high temperature when the value is large, and a low temperature when the value is low. If the luminance value is large, the temperature may be high, and if the value is low, the temperature may be low.

1 is a flowchart illustrating a temperature correction method according to a first embodiment.

Referring to FIG. 1 , the thermal image may be input (S10). The three fpa values may be input (S11).

The thermal image and the fpa value may be input together in a single step. By using the luminance of the thermal image and the three fpa values, an fpa value of the thermal image may be obtained. Specifically, if the luminance of the current pixel of the thermal image is higher than the luminance value of the center of the thermal image, the step S121 of applying the first conversion equation is performed. If the luminance of the current pixel of the thermal image is lower than the luminance value of the center of the thermal image, the process proceeds to the step S122 of applying the second conversion equation. The first transformation equation may be expressed as Equation 1, and the second transformation equation may be expressed as Equation 2.

Using Equations 1 and 2, fpa values of all pixels of the thermal image can be obtained (S13). This can be called an fpa value map. The above equation can scale the luminance scale of the entire screen and the scale of the fpa value in both directions based on the fpa value of the center of the image. Here, any one of the two directions is the direction of Equation 1, and the other direction of the two directions is the direction of Equation 2 above.

According to the conversion equation, the fpa value of the body may be further reflected according to the possibility that the body to be measured is placed in the center of the image.

According to the two conversion equations, the fpa values of all pixels are obtained in both directions using the luminance as reference information. Accordingly, it is possible to eliminate the problem of white balance of the thermal image. The white balance refers to a phenomenon in which the radiant energy of the surrounding area is greatly conspicuous when it is close to a hot material. For example, let's take an example of a body in the center of the screen (36 degrees Celsius) holding coffee at a fairly hot temperature (80 degrees Celsius) in one corner of the screen. In this case, the influence of the pipe balance by the coffee may be weakened.

According to the two conversion equations, the most accurate map of fpa values can be obtained by using all three fpa values and luminance information.

The general image may be input (S20), and together with it, location information of a hot spot and temperature information of the location among the general images may be input (S21). Here, the temperature information may be input by directly measuring it by a device providing ground truth. The location information may be the forehead, but may also be the forehead in the body. It is good to be in a hot place as much as possible, but it is not limited to a specific location. The general image input ( S20 ) and the pair input step ( S21 ) of location information and temperature information will be described in more detail by step S42 of FIG. 3 .

Thereafter, the temperature information in the location information in the body and the fpa value of the pixel of the thermal image corresponding to the location information may be correlated with each other ( S30 ). In this case, it is possible even if it is not the highest fpa value. However, in order to improve the accuracy of the scaling and reduce the complexity of the calculation, it is preferable to use the location information and the temperature information of the place with the highest temperature in the body in the step of associating the fpa ( S30 ). For example, the temperature information of the general image may be measured at a position having the highest fpa value.

It is sufficient that the step (S30) of associating the fpa value is performed only once when the device starts to operate. The step of associating the fpa value (S30) and the previous step do not need to be performed again while the device related to the present invention continues to operate. Of course, there is no need for the user to measure the ground truth body temperature as a new sample again. For example, when the body temperature is to be measured next time, the body temperature may be found by using the fpa value obtained from the thermal image and the corresponding temperature value. Of course, temperature distribution information may be output using the fpa value map.

The body may be the face of the subject to be measured.

After that, it is possible to find out the temperature distribution of all the body by using a predetermined proportional relationship, that is, the proportional relationship of the fpa value (S31). A point having the highest temperature among the temperature distribution may be output as the body temperature.

FIG. 2 is a view schematically showing the process of FIG. 1 .

Referring to FIG. 2 , the rearmost image is Tij, indicating thermal image information. The middle image shows the fpa value map of the thermal image using the fpa values of three points. It can be seen that the first image shows that the thermal image is a temperature image, Tempij.

By the method shown in FIGS. 1 and 2 , variables such as external temperature and humidity and camera internal temperature can act independently. This is because one calibration was performed at the start of the operation. However, the inaccuracy according to the distance between the body and the camera was not corrected. A second embodiment for improving this will be described with reference to FIGS. 3 and 4 .

FIG. 4 is a diagram showing a rough correlation between distance information between a camera and a body, the size of the box image, and an fpaM value.

Referring to FIG. 4 , the size of the box image decreases as the distance increases, and the fpaM value also decreases in general. This can occur in instruments that measure radiant energy.

3 is a flowchart illustrating a temperature correction method according to the second embodiment.

Referring to FIG. 3 , a correction value δ according to distance may be obtained using a plurality of the general images. For example, various images are input (S40), the distance d between the camera and the body is input from the input image (S41), and the box image is extracted (S42). To the box image, a predetermined extraction algorithm for extracting a specific part of the body from the entire input image may be applied. For example, an algorithm for extracting faces known as Retinaface can be used. The box image may be extracted as a rectangle (boxsize) of a predetermined size (S43). The extracted information may be stored in a memory (S44).

Separately from the operation, the thermal image may be input (S50), and the highest fpa value may be extracted from the thermal image and stored in the memory. The highest fpa value can be obtained in the body temperature distribution acquisition step (S31) of FIG. Of course, if there is no noise such as a high-temperature object, and if it is the same person's body, the simple highest fpa value input from the thermal imaging camera may be used instead of the body temperature distribution acquisition step ( S31 ).

In the memory, information about the thermal image and the general image measured at various distances may be recorded. For example, hundreds of data can be obtained at distances d of 0.5 meters, 1 meter, 2 meters, and 3 meters, respectively. Among the data, the box size may use an average value of a plurality of experiments. In this case, the thermal image and the general image may be information obtained under the same condition.

The correction value δ of the fpa value may be calculated using the information stored in the memory (S45).

First, Equation 3 can be used to correlate the distance and the box size. In fact, since the distance is difficult to obtain, the fpa value is corrected using the box size.

Here, d hat is the estimated distance, mu is the average size value of the box, the number indicates the distance step, and α may be a proportional constant that fits the distance from each distance step to the next step.

According to Equation 3, given the size of the box image, an estimated value of the distance can be obtained. Thereafter, a correction value (δ) of the fpa value according to the estimated distance can be obtained. Specifically, it can be obtained as a value obtained by multiplying the estimated distance by subtracting the distance of the previous step and the value obtained by subtracting the fpa value of the previous step from the fpa value of the subsequent step. For example, when the estimated distance d-hat is between 1-2 meters, Equation 4 may be applied. In this case, the highest value of fpa may be used. Here, other values may be used for the fpa value, but it is preferable because the highest body temperature can identify the body temperature that selects the patient. In addition, it is preferable to use the highest value because accuracy can be improved at the time of scaling.

When the estimated distance d-hat exceeds a certain distance, the fpa value hardly changes, so there is no need for correction. For example, when the estimated distance d-hat is 3 meters or more, the correction value (δ) of the fpa value can be used in the same way.

Through the above steps, the highest fpaM value can be obtained using the correction value δ of the fpa value according to the size of the box image (S52).

Thereafter, distance-corrected body temperature distribution information can be obtained (S53).

Here, the body may mean a face.

According to the above embodiment, since it can be implemented with a simple algorithm and the amount of calculation is large, the body temperature can be measured only with a microcomputer.

According to the present embodiment, it is possible to measure the body temperature more accurately by correcting an error according to the distance between the body and the camera.

According to the present invention, the body temperature can be measured without the need for expensive equipment and without the influence of various sources of error (temperature, humidity, and distance).

Claims (9)

obtaining an fpa value map of the thermal image using the thermal image and fpa values of at least three pixels of the thermal image;
associating a general image of a specific point whose temperature value is known and an fpa value of the thermal image corresponding to the specific point; and
It includes obtaining temperature distribution information or body temperature information of a thermal image using the temperature value of the specific point and the fpa value map,
Temperature information correction method for body temperature measurement. The method of claim 1,
At least one of the fpa values is a value of the center of the thermal image. The method of claim 1,
At least one of the fpa values is the highest value in the thermal image, and at least one of the fpa values is the lowest value in the thermal image. The method of claim 1,
The specific point is a place where the temperature is high in the thermal image, temperature information correction method for body temperature measurement. The method of claim 1,
The fpa value map is obtained by Equations 1 and 2,
Equation 1:

Equation 2:

Here, fpa means fpa value, i and j mean the x and y-axis index of the pixel of the thermal image, gray means the luminance of the black-and-white image that is the thermal image, T means the luminance, and M is means max, m means min, c means center,
Temperature information correction method for body temperature measurement. The method of claim 1,
In order to correct an error according to the distance at which the thermal image is acquired, correcting the highest fpa value in the thermal image is further performed
Temperature information correction method for body temperature measurement. 7. The method of claim 6,
wherein the highest fpa value is an fpa value of a pixel having the highest temperature among the temperature distribution information. 7. The method of claim 6,
The correction is performed by adding a certain correction value,
Temperature information correction method for body temperature measurement. 9. The method of claim 8,
Any of the above correction values,
A method for correcting temperature information for body temperature measurement, wherein the size of the box extracted by an extraction algorithm for extracting a part of the image is obtained by using that the size of the extracted box becomes smaller according to the distance at which the thermal image is acquired.

Images