DISEASE DETECTION DEVICE AND DISEASE DETECTION SYSTEM

Abstract

To provide a patient determination device capable of appropriately determining whether a person to be measured is a patient or a healthy person, the person to be measured having different exposed parts depending on presence / absence of a wearing article.SOLUTION: A patient determination device 2 comprises: a temperature pattern calculation unit 22 that calculates a temperature pattern at an exposed part of a person to be measured on the basis of temperature data input from a body surface temperature measurement device 1 for measuring a body surface temperature of the person to be measured and a wearing article determination result showing determined presence / absence of a wearing article of the person to be measured; and a person-to-be-measured determination unit 24 that applies one learned model selected from a plurality of learned models on the basis of the wearing article determination result to the temperature pattern to determine whether the person to be measured is a patient or a healthy person.SELECTED DRAWING: Figure 3

Description

[Technical field]

[0001] The present invention relates to a diseased person discrimination device and a diseased person discrimination system. [Background technology]

[0002] Conventionally, in many facilities, body surface temperature of subjects is measured periodically. Subjects include, for example, inpatients and outpatients in hospitals, passengers at airports, visitors to buildings, and children at educational facilities. In recent years, thermography devices have been used more and more to measure the body surface temperature of subjects without contact. Thermography devices can display the body surface temperature of subjects by thermal images that are color-coded, for example, red for high temperature and blue for low temperature, depending on the intensity of infrared light emitted from the skin of the subjects.

[0003] When a person to be measured is infected with a virus and develops symptoms, the person's body surface temperature rises. For this reason, a thermal image of the person to be measured is taken with a thermography device before entering a facility to distinguish between healthy and sick people. A predetermined temperature threshold has been used to distinguish between sick and healthy people using the thermal image displayed on the thermography device. For example, a person with a part of the face with a temperature higher than the temperature threshold is judged as a sick person, and a person without a part of the face with a temperature higher than the temperature threshold is judged as a healthy person.

[0004] Also, various methods have been attempted to distinguish between diseased and healthy people. For example, Patent Document 1 discloses a technique for setting a criterion for judging the condition of a plurality of subjects from their temperatures, and judging the condition of the subject from the temperature of the subject identified by a temperature identification means based on the criterion. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] JP 2020-62198 A Summary of the Invention [Problem to be solved by the invention]

[0006] In recent years, it has become common to wear face masks regardless of the season as a measure against infectious diseases. In addition, even people who do not normally wear glasses are increasingly wearing anti-pollen glasses with hoods on the rim during hay fever season to prevent pollen from getting into their eyes. However, when part of the face is covered, the exposed part of the face that can be measured by the thermography device is reduced. For this reason, conventional discrimination devices that use the detection results of a thermography device to determine whether a person is healthy or sick may erroneously determine that a person being measured is healthy even though he or she is a sick person with a fever.

[0007] In addition, the criteria used to determine whether a person is ill or healthy in the technology disclosed in Patent Document 1 must be adjusted to an appropriate value according to the environment, and if the criteria are not adjusted, the probability of misjudging the person being measured increases. In order to adjust the criteria to an appropriate value, the adjuster needs experience, but if the environment changes, the criteria must be readjusted, which has been an operational problem. Even if the technology disclosed in Patent Document 1 is used, for example, if part of the face of the person being measured is covered by a mask or the like, the body surface temperature of the person being ill cannot be measured correctly, and the person being measured who is actually ill may be misjudged as a healthy person.

[0008] The present invention has been made in consideration of such circumstances, and aims to correctly determine whether a person being measured is a patient or a healthy person, even if part of the person's exposed area is covered by clothing. [Means for solving the problem]

[0009] In order to achieve the above object, a disease discrimination device according to one aspect of the present invention is a device for determining a temperature pattern of a temperature in an exposed part of a subject based on temperature data input from a body surface temperature measuring unit that measures the body surface temperature of the subject and a clothing determination result that determines whether or not the subject is wearing clothing.When the result of the wear determination is that the wear is present, the temperature pattern is represented by a group of temperature data collected from the exposed parts of the subject other than the part where the wear is present, and when the result of the wear determination is that the wear is not present, the temperature pattern is represented by a group of temperature data collected from the exposed parts of the subject. A temperature pattern calculation unit that calculates a temperature pattern based on the result of the wear The results of the clothing judgment are different when the person is wearing clothing and when the person is not wearing clothing. The device is also equipped with a subject discrimination unit that applies one learned model selected from the multiple learned models to a temperature pattern to determine whether the subject is a diseased or healthy person. In addition, a disease discrimination system reflecting another aspect of the present invention includes, in addition to the above-mentioned disease discrimination device, a learning server having a temperature pattern learning unit that learns a temperature pattern corresponding to the worn clothing assessment result using a machine learning model and generates a plurality of learned models corresponding to the worn clothing assessment result. Effect of the Invention

[0010] According to the present invention, by applying a learned model selected based on the wearing article judgment result to the temperature pattern of the person being measured, it is possible to correctly determine whether the person being measured is a sick person or a healthy person. Problems, configurations and effects other than those described above will become apparent from the following description of the embodiments. [Brief description of the drawings]

[0011] [Figure 1] 1 is a block diagram showing an example of the internal configuration of a disease discrimination system according to a first embodiment of the present invention. [Diagram 2] 2 is a block diagram showing an example of the internal configuration of a control unit according to the first embodiment of the present invention; FIG. [Diagram 3] 1 is a block diagram showing an example of the internal configuration of a disease discrimination device according to a first embodiment of the present invention. [Figure 4] 1 is a block diagram showing an example of a hardware configuration of a computer according to a first embodiment of the present invention. [Diagram 5] 1 is a table showing an example of a discrimination result in which a subject is discriminated between a healthy person and a diseased person by applying a machine learning model according to the first embodiment of the present invention. [Figure 6] FIG. 11 is a block diagram showing an example of the internal configuration of a disease discriminating apparatus according to a second embodiment of the present invention. [Figure 7]FIG. 13 is a block diagram showing an example of the internal configuration of a disease discriminating system according to a third embodiment of the present invention. [Figure 8] FIG. 13 is a block diagram showing an example of the internal configuration of a disease discriminating apparatus according to a third embodiment of the present invention. [Figure 9] FIG. 13 is a block diagram showing an example of the internal configuration of a disease discrimination device according to a fourth embodiment of the present invention. [Figure 10] FIG. 13 is a block diagram showing an example of the internal configuration of a disease discrimination system according to a fifth embodiment of the present invention. [Figure 11] FIG. 13 is a block diagram showing an example of the internal configuration of a control unit according to a fifth embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0012] Hereinafter, an embodiment of the present invention will be described with reference to the accompanying drawings. In this specification and the drawings, components having substantially the same functions or configurations are designated by the same reference numerals, and redundant description will be omitted.

[0013] FIG. 1 is a block diagram showing an example of the internal configuration of a disease patient discrimination system 100. As shown in FIG. The diseased person discrimination system 100 includes a body surface temperature measuring device 1 and a diseased person discrimination device 2. The body surface temperature measuring device 1 and the diseased person discrimination device 2 are connected by wire or wirelessly. The diseased person discrimination device 2 captures a thermal image representing the body surface temperature of the subject measured by the body surface temperature measuring device 1, and discriminates whether the subject is a diseased person or a healthy person. A detailed configuration example of the diseased person discrimination device 2 will be described later with reference to FIG. 3.

[0014] The body surface temperature measuring device 1 (an example of a body surface temperature measuring section) includes a lens 10, a detection element 11, an A / D conversion section 12, a control section 13, a display section 14, and an operation section 15. This body surface temperature measuring device 1 is capable of measuring the body surface temperature of a person to be measured, to which the lens 10 is directed.

[0015] Lens 10 collects infrared light emitted from the person being measured and forms an image on detection element 11 . The detection element 11 has one or more sensors 11a, and outputs a detection signal in a time-division manner for each frame period in synchronization with a vertical synchronization signal input to the detection element 11. The number of frames of the detection signal that the detection element 11 can output per unit time is, for example, 60 fps (frames per second).

[0016] When the body surface temperature measuring device 1 is used as a camera, the sensor 11a of the detection element 11 is a sensor provided in pixel units. The sensor 11a detects the intensity of the infrared light imaged on the detection element 11 and outputs a detection signal. For example, a microbolometer whose resistance value changes according to the absolute temperature of the object to be measured is used as the sensor 11a. In the detection element 11, a large number of microbolometers are arranged two-dimensionally, for example, 320 microbolometers horizontally by 240 microbolometers vertically. When the microbolometer detects the infrared light, it outputs an analog detection signal according to the intensity of the infrared light. However, only one sensor 11a may be used in the detection element 11.

[0017] The A / D conversion unit 12 converts the detection signal output from the pixel sensor 11 a of the detection element 11 into digital data, and outputs the digital data to the control unit 13 .

[0018] The control unit 13 performs gain correction and offset correction shown in Fig. 2, which will be described later, on the digital data of the detection signal input from the A / D conversion unit 12, and calculates the absolute temperature of the object to be measured. Then, the control unit 13 outputs image data obtained by performing image processing based on the calculated absolute temperature of the object to the display unit 14. Here, the image data output by the control unit 13 is image data of a thermal image generated in accordance with the operation of the operation unit 15. The control unit 13 also outputs a vertical synchronization signal to the sensor 11a of the detection element 11.

[0019] The control unit 13 also outputs temperature data to the diseased person discrimination device 2. The temperature data includes a thermal image of the entire face of the subject including specific parts of the face, or the temperature of the specific parts measured for each specific part. The specific parts are, for example, at least one of the following: the inner corners of the eyes, the tip of the nose, the sides of the nose, the cheeks, the forehead, the temples, the neck, and the hands of the subject. If the face of the subject is not covered by clothing, any part can be selected from the specific parts described above. Note that the specific parts of the subject are, for example, parts that are not covered by clothing (jacket, shirt, pants, socks, hat, etc.) or footwear (shoes, sandals, etc.) and are not covered by clothing. In this embodiment, the clothing of the subject includes, for example, at least one of a mask, glasses, collar, scarf, and gloves. It is assumed that the positions of the specific parts to be extracted from the thermal image of the face by the specific part determination unit 21 shown in FIG. 3 described later are determined in advance. However, if the subject's face is covered by clothing, the aforementioned specific parts may be excluded from the specific parts, including the corners of the eyes, the tip of the nose, the sides of the nose, and the cheeks, which are likely to be covered by a mask or glasses. In addition, the neck, which is likely to be covered by clothing such as a collar or scarf, and the hands, which are likely to be covered by gloves, may also be excluded from the specific parts.

[0020] On the basis of the image data generated by the control unit 13, the display unit 14 displays a thermal image showing the distribution of the absolute temperature of the object to be measured.

[0021] The operation unit 15 is used to input an instruction to output the absolute temperature of the object to be measured to the control unit 13. In response to an instruction from the operation unit 15, the control unit 13 outputs image data to be displayed as a thermal image to the display unit 14.

[0022] Note that the display unit 14 may be removed from the configuration of the body surface temperature measuring device 1 by connecting a PC (Personal Computer) terminal or the like to the body surface temperature measuring device 1. In this case, the control unit 13 outputs to the PC terminal the absolute temperature of the object to be measured, calculated based on the detection signal input from the detection element 11 for each frame. This makes it possible to display the absolute temperature of the object to be measured on the PC terminal or the like connected to the body surface temperature measuring device 1.

[0023] FIG. 2 is a block diagram showing an example of the internal configuration of the control unit 13. As shown in FIG. The control unit 13 includes a gain correction unit 13a, an offset correction unit 13b, a calculation unit 13c, a thermal image generation unit 13d, and a worn item determination unit 13e.

[0024] The detection signal output from each sensor 11a of the detection element 11 is expressed by approximating a linear equation that has a larger value as the absolute temperature of the measurement object increases. However, the characteristics of the microbolometers corresponding to each sensor 11a of the detection element 11 vary, and the linear equation itself deviates from a standard linear equation. For this reason, for example, when the body surface temperature measuring device 1 is shipped from the factory, correction values ​​used for gain correction and offset correction for aligning the characteristics of the microbolometers corresponding to each sensor 11a are stored in a memory (not shown). Then, the control unit 13 corrects the digital data based on the correction values ​​read from the memory.

[0025] The gain correction unit 13a of the control unit 13 performs gain correction on the detection signal converted into digital data input from the A / D conversion unit 12. The gain correction unit 13a corrects the gain, that is, the slope of the linear expression. The offset correction unit 13b performs offset correction on the digital data whose gain has been corrected by the gain correction unit 13a. The offset correction unit 13b corrects the offset, that is, the intercept of the linear expression.

[0026] The calculation unit 13c calculates the absolute temperature of the object to be measured for each pixel based on the digital data obtained by converting the detection signals corrected by the gain correction unit 13a and the offset correction unit 13b. The thermal image generating unit 13d generates image data for displaying a thermal image based on the absolute temperature of the measurement object calculated for each sensor 11a by the calculating unit 13c. The image data generated by the thermal image generating unit 13d is output to the wearing article determining unit 13e and the display unit 14. The display unit 14 displays the image data input from the thermal image generating unit 13d as a thermal image of the measurement object.

[0027] The worn article determination unit 13e outputs a worn article determination result that determines whether or not the subject is wearing an article based on image data (thermal image) input from the thermal image generation unit 13d. For example, it determines that the subject is wearing an article at a part of the subject's face where there is a sudden temperature change. As described above, the worn article of the subject includes at least one of a mask, glasses, collar, muffler, and gloves. If the worn article determination unit 13e determines that the subject is not wearing any of a mask, glasses, collar, muffler, and gloves, it obtains a determination result that the subject is wearing no article. On the other hand, if the worn article determination unit 13e determines that the subject is wearing at least one of a mask, glasses, collar, muffler, and gloves, it obtains a determination result that the subject is wearing an article. The determination result of whether or not the subject is wearing an article is output to the disease discrimination device 2 as a worn article determination result. In this way, the worn article determination result is a result of determining whether or not the subject is wearing at least one of a mask, glasses, collar, muffler, and gloves. In the following figures, the input and output of the wearing article determination result may be indicated by dashed lines. In the following description, the wearing article will be described as a mask and glasses.

[0028] FIG. 3 is a block diagram showing an example of the internal configuration of the disease patient discrimination apparatus 2. As shown in FIG. As described above, the diseased person discrimination device 2 discriminates whether the subject is a diseased person or a healthy person based on the information acquired from the body surface temperature measuring device 1. The diseased person discrimination device 2 includes a specific part determination unit 21, a temperature pattern calculation unit 22, a memory unit 23, a subject discrimination unit 24, a display unit 25, an input unit 26, and a temperature pattern learning unit 27. The wearing article determination result input from the wearing article determination unit 13e of the body surface temperature measuring device 1 is used by the specific part determination unit 21, the temperature pattern calculation unit 22, the memory unit 23, and the temperature pattern learning unit 27.

[0029] The specific part determination unit 21 determines a specific part of the exposed part of the subject based on the clothing determination result and temperature data input from the body surface temperature measuring device 1. The temperature data includes a thermal image of the exposed part of the subject captured by the body surface temperature measuring device 1. In this embodiment, when the clothing determination result is a determination result of no clothing, the specific part of the subject includes at least one of the following: the inner corners of the eyes, the tip of the nose, the sides of the nose, the cheeks, the forehead, the temples, the neck, and the hands. When the clothing determination result is a determination result of the presence of clothing, the exposed part not covered by clothing includes, for example, the temples, the forehead, the neck, and the hands. These exposed parts are used as the specific parts of the subject.

[0030] The face of the person being measured is not necessarily facing the lens 10. For this reason, the specific part determination unit 21 has a function of aligning the facial parts that make up the face based on a thermal image of the face. The specific part determination unit 21 then uses the clothing determination result to align the facial parts that make up the face with respect to the thermal image and determine the specific part from the thermal image. Here, the clothing determination result input from the body surface temperature measuring device 1 is used so that the specific part determination unit 21 can select a facial part to be used as a clue for aligning the thermal image.

[0031] The specific part determination unit 21 is configured to output the thermal image after alignment to the temperature pattern calculation unit 22. However, the specific part determination unit 21 may be configured to output to the temperature pattern calculation unit 22 a combination of the thermal image before alignment and a group of coordinates of the specific part in the image, or the like.

[0032] The specific part determination unit 21 requires the wearing article determination result not only when outputting the aligned thermal image to the temperature pattern calculation unit 22, but also when outputting a coordinate group representing the position of the specific part determined in the thermal image to the temperature pattern calculation unit 22. For example, if the determination result is that no wearing article is present, the specific part determination unit 21 outputs a coordinate group of parts selected mainly from the face of the subject to the temperature pattern calculation unit 22. On the other hand, if the determination result is that wearing article is present, the specific part determination unit 21 outputs a coordinate group of parts excluding parts covered by wearing article to the temperature pattern calculation unit 22.

[0033] The temperature pattern calculation unit 22 calculates a temperature pattern of the temperature in the exposed part of the subject based on the temperature data input from the body surface temperature measuring device 1 that measures the body surface temperature of the subject and the result of the wearing article judgment that judges whether the subject is wearing an article. At this time, the temperature pattern calculation unit 22 extracts a group of temperature data from the aligned thermal image input from the specific part judgment unit 21 based on the group of coordinates input from the specific part judgment unit 21, and calculates a temperature pattern of the specific part based on the temperature in the exposed part of the subject. The temperature pattern is represented by a group of temperature data collected from a plurality of specific parts of the subject. In addition, the temperature pattern represents a group of temperature data collected from the exposed part of the subject other than the part where the article is worn when the wearing article judgment result is the presence of an article. In addition, the temperature pattern is represented by a group of temperature data collected from the exposed part of the subject when the wearing article judgment result is the absence of an article.

[0034] A specific part includes multiple pixels. Therefore, the temperature pattern calculation unit 22 calculates the average value of multiple pixels included in the area determined as the specific part as the temperature of the specific part. In other words, the temperature pattern calculation unit 22 extracts temperature data from the specific part determined by the specific part determination unit 21, and can obtain temperature patterns of multiple specific parts. At this time, the temperature pattern calculation unit 22 uses the worn article determination result input from the body surface temperature measuring device 1 to specify the position to be extracted from the input aligned thermal image.

[0035] The temperature pattern calculated by temperature pattern calculation unit 22 and determined by the wearing article judgment result is output to subject discrimination unit 24 and temperature pattern learning unit 27 in association with a plurality of specific parts. Temperature pattern calculation unit 22 may calculate the maximum value among the values ​​of a plurality of pixels included in the area judged as the specific part as the temperature of the specific part, and output the temperature pattern of the specific part to subject discrimination unit 24 and temperature pattern learning unit 27 in association with the calculated temperature.

[0036] The memory unit 23 stores a plurality of trained models in which temperature patterns have been learned according to the presence or absence of clothing worn by the subject, and the results of the clothing determination. The trained models are used by the subject discrimination unit 24 to discriminate the subject as either a sick person or a healthy person based on the temperature pattern of the subject. In this embodiment, the memory unit 23 stores the results of the clothing determination of the subject measured by the body surface temperature measuring device 1. The trained models stored in the memory unit 23 differ depending on the results of the clothing determination of the subject. For example, at least two types of trained models, a first trained model that is applied when the clothing determination result is no clothing and a second trained model that is applied when clothing is worn, are stored in the memory unit 23.

[0037] When the diseased person discrimination device 2 has a learning function, the learned model is one that has been learned by a temperature pattern learning unit 27 described later. However, when the diseased person discrimination device 2 does not have the learning function, the learned model is one that has been prepared in advance in the storage unit 23.

[0038] The subject discrimination unit 24 applies one learned model selected from a plurality of learned models to the temperature pattern based on the subject's clothing determination result read from the storage unit 23, and determines whether the subject is a sick person or a healthy person. At this time, the subject discrimination unit 24 discriminates the subject using the learned model read from the storage unit 23 based on the clothing determination result read from the storage unit 23.

[0039] Therefore, the subject discrimination unit 24 selects different learned models depending on whether the worn article determination result is that an article is being worn or that an article is not being worn. For example, the subject discrimination unit 24 selects and reads out from the storage unit 23 either the first learned model or the second learned model applicable to the temperature pattern calculated by the temperature pattern calculation unit 22, depending on the worn article determination result. When the worn article determination result is that an article is not being worn, the subject discrimination unit 24 reads out the first learned model from the storage unit 23 and applies the first learned model to the temperature pattern. On the other hand, when the worn article determination result is that an article is being worn, the subject discrimination unit 24 reads out the second learned model from the storage unit 23 and applies the second learned model to the temperature pattern.

[0040] Then, subject discrimination unit 24 applies the first or second learned model selected and read from storage unit 23 to the temperature pattern calculated by temperature pattern calculation unit 22 to discriminate whether the subject is a diseased person or a healthy person. The discrimination result by subject discrimination unit 24 is output to display unit 25.

[0041] The display unit 25 displays the result of the measurement subject discrimination unit 24 discriminating the measurement subject, for example, a discrimination result including the possibility that the measurement subject is a sick person. As one example, a text message indicating whether the measurement subject is a sick person or a healthy person is displayed on the display unit 25 as the discrimination result. Note that instead of the display unit 25, a notification unit that issues an alarm may be provided. The alarm is notified, for example, by the notification unit turning on a lamp or sending an email to the facility manager.

[0042] Next, a description will be given of an example of the operation of the input unit 26 and the temperature pattern learning unit 27, which are responsible for the process of generating a trained model. Note that the process of generating a trained model is not always performed. The input unit 26 accepts the user's discrimination result as to whether the person being measured is a healthy person or a patient. This discrimination result is input by the user himself using the patient discrimination device 2. For example, if the person being measured is actually a healthy person, the user inputs "healthy person" through the input unit 26. On the other hand, if the person being measured is actually a patient, the user inputs "patient" through the input unit 26.

[0043] The user inputs the user discrimination result through the input unit 26 and corrects the trained model. The input user discrimination result is either "healthy person" or "patient" as described above. When the user inputs the user discrimination result through the input unit 26, the user does not need to always check the output of the display unit 25 to determine whether the discrimination result of the subject discrimination unit 24 is correct or not.

[0044] The temperature pattern learning unit 27 also receives the worn item determination result from the control unit 13 of the body surface temperature measuring device 1. The temperature pattern learning unit 27 then corrects the first or second learned model by receiving the temperature pattern provided by the temperature pattern calculation unit 22, the user determination result input from the input unit 26 indicating the correct answer as to whether the temperature pattern should be determined as a healthy person or a sick person, and the worn item determination result of the subject.

[0045] For this reason, a plurality of learning models are prepared in the temperature pattern learning unit 27, and the learning model to be learned is selected based on the wearing article discrimination result input from the body surface temperature measuring device 1. That is, the temperature pattern learning unit 27 learns a first or second learned model learned by a machine learning model based on the user discrimination result input from the input unit 26, the wearing article discrimination result input from the body surface temperature measuring device 1, and the temperature pattern input from the temperature pattern calculation unit 22. The temperature pattern used here is the temperature pattern used by the measurement subject discrimination unit 24.

[0046] The first or second trained model is data that accumulates the learning results of the temperature pattern learning unit 27. In this way, the temperature pattern learning unit 27 generates at least two types of trained models according to the wearing article determination result. However, three or more types of trained models may be generated. In the following description, when there is no need to distinguish between the first and second trained models, they are collectively referred to as the "trained models."

[0047] The temperature pattern learning unit 27 updates the learned model by the above-mentioned machine learning and stores the updated learned model in the memory unit 23. The learning model is updated regardless of whether the answer is correct or incorrect. In other words, the learned model is corrected and updated regardless of whether the output of the subject discrimination unit 24 matches the actual situation or not. As the temperature pattern learning unit 27 repeats learning, the learned model stored in the memory unit 23 is gradually corrected and updated to an appropriate learned model. Thereafter, the subject discrimination unit 24 applies the updated learned model to the temperature pattern calculated by the temperature pattern calculation unit 22 to determine whether the subject is a sick person or a healthy person.

[0048] Next, the hardware configuration of the computer 30 constituting each device of the disease patient discrimination system 100 will be described. 4 is a block diagram showing an example of a hardware configuration of the computer 30. The computer 30 is used as hardware constituting the body surface temperature measuring device 1 and the diseased person discrimination device 2, for example.

[0049] The calculator 30 is hardware used as a so-called computer. The calculator 30 includes a CPU (Central Processing Unit) 31, a ROM (Read Only Memory) 32, and a RAM (Random Access Memory) 33, each of which is connected to a bus 34. The calculator 30 further includes a display device 35, an input device 36, a non-volatile storage 37, and a network interface 38.

[0050] The CPU 31 reads out program code of software for realizing each function according to this embodiment from the ROM 32, loads it into the RAM 33, and executes it. Variables, parameters, etc. generated during the calculation processing of the CPU 31 are temporarily written to the RAM 33, and these variables, parameters, etc. are read out appropriately by the CPU 31. In the body surface temperature measuring device 1, the function of the control unit 13 is mainly realized by the CPU 31. In the disease discrimination device 2, the functions of the specific part determination unit 21, the temperature pattern calculation unit 22, the measurement subject discrimination unit 24, and the temperature pattern learning unit 27 are realized by the CPU 31.

[0051] The display device 35 is, for example, a liquid crystal display monitor, and displays to the user the results of the processing performed by the computer 30, such as a thermal image. For example, the display device 35 is used for the display unit 14 of the body surface temperature measuring device 1 and the display unit 25 of the diseased person discrimination device 2.

[0052] The input device 36 may include, for example, an operation key, an operation button, etc., and allows the user to input predetermined operations and give instructions. For example, the input device 36 is used in the operation unit 15 of the body surface temperature measuring device 1 and the input unit 26 of the diseased person discrimination device 2.

[0053] As the nonvolatile storage 37, for example, a hard disk drive (HDD), a solid state drive (SSD), a flexible disk, an optical disk, a magneto-optical disk, a CD-ROM, a CD-R, a magnetic tape, a nonvolatile memory, etc. are used. In addition to an operating system (OS) and various parameters, a program for making the computer 30 function is recorded in this nonvolatile storage 37. The ROM 32 and the nonvolatile storage 37 permanently record programs and data necessary for the CPU 31 to operate, and are used as an example of a computer-readable non-transient recording medium storing a program executed by the computer 30. For example, the nonvolatile storage 37 is used in the memory unit 23 of the diseased person discrimination device 2.

[0054] For example, a NIC (Network Interface Card) or the like is used as the network interface 38, and various data can be transmitted and received between the devices via a LAN (Local Area Network) connected to a terminal of the NIC, a dedicated line, etc. For example, the network interface 38 is used in a communication unit (not shown) that connects the body surface temperature measuring device 1 and the diseased person discrimination device 2 so that they can communicate with each other.

[0055] FIG. 5 is a list showing an example of the discrimination result in which a machine learning model is applied to discriminate subjects into healthy subjects and subjects with a disease. In FIG. 5, the user discrimination result input by the user in advance is expressed as the "teacher label (input)." The discrimination result of the subject output by subject discrimination unit 24 on display unit 25 after applying the learned model to the temperature pattern is expressed as the "discrimination result (output)." The "teacher label (input)" and "discrimination result (output)" items each include the items "healthy subject" and "disease subject."

[0056] In conventional disease identification systems using infrared cameras, an alarm would be issued if the maximum temperature value in a thermal image of a face exceeded a temperature threshold. However, there is a large individual difference in body surface temperature, and if a fixed temperature threshold was used, healthy people could be mistakenly identified as diseased, causing frequent alarms and confusion at the operation site. In order to avoid this, if the temperature threshold was raised too high, there was a problem that diseased people could not be screened and would pass through. Furthermore, body surface temperature is easily affected by the outside temperature, and users had to change the temperature threshold to an appropriate value depending on the outside temperature, which was time-consuming.

[0057] Here, when the method according to the first embodiment is used, the setting of a temperature threshold, which has been conventionally performed, becomes unnecessary. For example, Table (1) in FIG. 5 is a list showing an example of the result of discriminating a subject not wearing glasses and a mask using the first trained model trained with a temperature pattern collected from the entire face. As shown in Table (1), 98.5% of healthy subjects were correctly discriminated as healthy subjects, but 1.5% of healthy subjects were erroneously discriminated as diseased subjects. Also, 96.6% of diseased subjects were correctly discriminated as diseased subjects, but 3.4% of diseased subjects were erroneously discriminated as healthy subjects. When the subject was not wearing any clothing and the entire face of the subject was exposed, generally good discrimination results were obtained.

[0058] Table (2) in FIG. 5 is a list showing an example of the results of discriminating subjects wearing glasses and a mask using the first trained model trained with temperature patterns collected from the entire face. As described above, the first trained model is obtained by machine learning using temperature patterns collected from the entire face. For this reason, 100.0% of healthy subjects were correctly discriminated as healthy subjects and were not mistakenly discriminated as diseased subjects. However, 100.0% of diseased subjects were also mistakenly discriminated as healthy subjects and were not correctly discriminated as diseased subjects. The reason for this is, for example, that the part of the subject wearing a mask does not correctly indicate the temperature of the face. In addition, glasses are difficult to transmit infrared rays of body surface temperature and are easily affected by the outside air, so the temperature of the part of the subject wearing glasses is likely to deviate from the body temperature of the subject.

[0059] When the rate of erroneous determination of diseased persons increases in this way, it becomes impossible to correctly identify a person as a diseased person, even if the person being measured is a diseased person. This is because, although the first trained model uses temperature patterns collected from the entire face to distinguish between healthy and diseased persons, the correct temperature pattern cannot be obtained because part of the face of the person being measured is covered.

[0060] For this reason, the first trained model, which was machine-learned using the temperature pattern collected from the entire face, could not accurately determine whether the subject was a sick person, especially when the subject was wearing a mask and glasses. Therefore, if the disease discrimination device 2 had a high rate of determining that the subject was a healthy person, even if the subject was actually a sick person, the sick person who was determined to be a healthy person may enter the facility.

[0061] Table (3) in FIG. 5 is a list showing an example of the results of discriminating subjects wearing glasses and a mask using the second trained model trained on temperature patterns collected from areas other than the mask and glasses. When the same experimental data as shown in Table (2) in FIG. 5 is used, the misclassification rate of healthy subjects is 1.0%. On the other hand, the disease discrimination rate is 93.1%, and the disease discrimination rate is 6.9%. In other words, both the healthy subject discrimination rate and the disease discrimination rate can achieve good values ​​of 90% or more. In this way, it was found that by applying an appropriate second trained model to a subject wearing clothing, the disease misclassification rate is reduced compared to the case where the first trained model applied to the case where no clothing is worn is applied. Therefore, by applying the first trained model when the discrimination result is no clothing and applying the second trained model when the discrimination result is the presence of clothing, superior results were obtained compared to the case where the same first trained model is applied to discriminate healthy subjects from diseased subjects regardless of the presence or absence of clothing.

[0062] In this way, the method according to the present embodiment makes it possible to generally distinguish between healthy people and people with diseases, regardless of whether the person is wearing clothing. Therefore, the disease discrimination device 2 can reliably determine that the person being measured is a person with a disease and restrict entry into the facility.

[0063] Table (3) in Figure 5 shows an example of the discrimination results for a subject wearing both a mask and glasses. However, results similar to those in Table (3) in Figure 5 can be obtained for subjects wearing only a mask or only glasses.

[0064] In the disease discrimination system 100 according to the first embodiment described above, by focusing on the difference in the facial temperature patterns between diseased and healthy people, it is possible to discriminate between diseased and healthy people using a trained model trained by a machine learning model. Then, the trained model is selected and applied according to whether the subject is wearing a mask and glasses. Therefore, it is possible to accurately discriminate between diseased and healthy people, regardless of whether the subject is wearing clothing on their face or not.

[0065] Here, the temperature patterns of the faces of sick people and healthy people are different. The temperature pattern is represented by a group of temperature data collected from multiple specific parts. Therefore, the sick person discrimination device 2 calculates a temperature pattern from the temperature data of each specific part measured by the body surface temperature measuring device 1, and can easily and correctly distinguish between sick people and healthy people by applying a pre-trained model. In addition, even if the inner corners of the eyes are hidden by the eyeglass frames, or the tip of the nose, sides of the nose, cheeks, etc. are hidden by a mask, the temperature pattern is calculated based on the exposed parts that are not covered by the eyeglasses and mask. Therefore, by applying the second trained model, it becomes possible to easily and correctly distinguish between sick people and healthy people.

[0066] When the subject is determined to be wearing clothing, the target of the temperature pattern to which the second trained model is applied is not limited to the face, but may be any exposed part of the human body. For example, the target of the temperature pattern may be the neck, the top of the head, the hands, etc.

[0067] Compared to conventional methods of distinguishing between sick and healthy people using a temperature threshold, the disease discrimination device 2 according to this embodiment applies a learned model to the calculated temperature pattern, allowing the disease discrimination device 2 to distinguish between sick and healthy people regardless of individual differences in the body surface temperature of the person being measured or the outside air temperature of the place where the person being measured was staying, reducing misidentification of sick people. Also, whereas it was difficult to set an appropriate temperature threshold depending on the gender and age of the person being measured, this embodiment utilizes the difference in temperature patterns between sick and healthy people, making it unnecessary to set delicate settings such as a temperature threshold, and facilitating operation.

[0068] Moreover, by learning temperature patterns using a machine learning model and using the accumulated learning results as a trained model, it becomes possible to distinguish between sick people and healthy people caused by various diseases. It is also possible to generate different trained models depending on whether the person being measured is a child or an adult, the gender, the nationality, etc. Furthermore, it is also possible to generate different trained models depending on the conditions of the operating location, such as the temperature. This improves the accuracy with which the patient discrimination device 2 distinguishes between sick people and healthy people.

[0069] Furthermore, even before onset of symptoms, i.e., before the subject's body surface temperature rises, the temperature change indicated by the temperature pattern is likely to show the same tendency as after onset of symptoms. Therefore, the disease discrimination device 2 can apply the learned model to the temperature pattern to discriminate even a pre-symptomatic subject as a diseased person. In this way, the disease discrimination device 2 can determine with a high probability that a subject is a diseased person. Furthermore, the disease discrimination device 2 can display on the display unit 25 the possibility that a pre-symptomatic subject is a diseased person, to alert the user.

[0070] In addition, multiple exposed parts of the subject are determined as specific parts, and the relative changes in temperature at the specific parts of the patient and healthy subjects are used as temperature patterns. The temperature patterns of the faces of the patient and healthy subjects are updated by the temperature pattern learning unit 27. Therefore, the trained model is updated by the temperature patterns related to the selected specific parts. Thus, the patient discrimination device 2 can use the updated trained model to discriminate patients caused by various infectious diseases, poor physical condition, etc.

[0071] Furthermore, the diseased person discrimination device 2 may also use a process of discriminating a subject whose body surface temperature is equal to or higher than a temperature threshold as a diseased person, as in the conventional case. Furthermore, the temperature pattern learning unit 27 may not only learn the relationship between the body surface temperatures of each specific part, but may also learn the body surface temperature of the subject itself. Therefore, the temperature pattern learning unit 27 can learn the absolute value of the body surface temperature of the subject. The subject discrimination unit 24 can also use the absolute value of the body surface temperature as a criterion for discriminating whether the subject is healthy or sick.

[0072] [Second embodiment] Next, a disease discriminating apparatus according to a second embodiment of the present invention will be described with reference to FIG.

[0073] FIG. 6 is a block diagram showing an example of the internal configuration of the disease patient discrimination device 2A. The diseased person discrimination device 2A is an integrated combination of the functional units of the body surface temperature measuring device 1 shown in Fig. 1 and the functional units of the diseased person discrimination device 2 shown in Fig. 3. Here, the lens 10, the detection element 11 and the sensor 11a, the A / D conversion unit 12, and the control unit 13 are collectively referred to as a body surface temperature measuring unit 16. The control unit 13 of the body surface temperature measuring unit 16 outputs temperature data to a specific part determination unit 21. The control unit 13 also outputs a result of the measurement subject's clothing determination to the specific part determination unit 21, the temperature pattern calculation unit 22, the storage unit 23, the measurement subject discrimination unit 24, and the temperature pattern learning unit 27. The processing of the specific part determination unit 21, the temperature pattern calculation unit 22, and the measurement subject discrimination unit 24 is the same as that shown in the first embodiment.

[0074] The diseased person discrimination apparatus 2A includes a display unit 14A and an operation unit 15A. Like the display unit 14 according to the first embodiment, the display unit 14A displays a thermal image based on image data output from the control unit 13. Also, like the display unit 25 according to the first embodiment, the display unit 14A can display the discrimination result of the subject or the possibility that the subject has a disease.

[0075] Like the operation unit 15 according to the first embodiment, the operation unit 15A inputs an instruction to the control unit 13 to output the absolute temperature of the object to be measured. Like the input unit 26 according to the first embodiment, the operation unit 15A can also receive a discrimination result as to whether the subject is a healthy person or a diseased person, and output the discrimination result to the temperature pattern learning unit 27. Note that this discrimination result is input by the user using the diseased person discrimination device 2A after checking the content displayed on the display unit 14A.

[0076] The diseased person discrimination device 2A according to the second embodiment described above can process the thermal image, determine whether the subject is wearing an article, and discriminate the subject, all in one device. This makes the diseased person discrimination device 2A lighter, smaller, and easier to carry.

[0077] [Third embodiment] Next, a disease patient discrimination system according to a third embodiment of the present invention will be described with reference to FIGS.

[0078] A body surface temperature measuring device 1 equipped with a display unit 14 capable of displaying thermal images is very expensive. However, a technology for distinguishing between sick and healthy individuals using temperature patterns can be used in a variety of environments, and it is therefore required that the body surface temperature measuring device 1 be inexpensive and easy to carry. Therefore, the sick individual discrimination system according to this embodiment is configured to be able to distinguish between sick and healthy individuals even when using a non-contact body surface temperature measuring device that does not have a display unit.

[0079] FIG. 7 is a block diagram showing an example of the internal configuration of the disease patient discrimination system 100A. The diseased person discrimination system 100A includes a body surface temperature measuring device 1A, a diseased person discrimination device 2B, and a learning server 7.

[0080] The body surface temperature measuring device 1A has a configuration in which the lens 10 and the display unit 14 are removed from the body surface temperature measuring device 1 shown in Fig. 1. The detection element 11 has one sensor 11a. For example, by configuring the external shape of the body surface temperature measuring device 1A to be pen-shaped, the user can bring the body surface temperature measuring device 1A close to a specific part of the person to be measured and measure the temperature of the specific part. It is assumed that the user, the body surface temperature measuring device 1A, the disease discrimination device 2B, and the learning server 7 are aware of the predetermined measurement order of the specific parts.

[0081] Then, the user discrimination result is input from the operation unit 15 to the control unit 13. As described above, the user discrimination result is a result of the user discriminating whether the subject is actually a healthy person or a diseased person.

[0082] Furthermore, the operation unit 15 may be an input button provided on the body surface temperature measuring device 1A. For example, the user checks whether the subject is wearing an article while looking at the subject, and presses an input button corresponding to the presence or absence of the article. The operation unit 15 outputs to the control unit 13 a result of the article determination corresponding to the pressed input button.

[0083] The control unit 13 outputs temperature data including temperatures measured by the body surface temperature measuring device 1A from specific parts of the subject in a predetermined measurement order using the body surface temperature measuring device 1A to the disease subject discrimination device 2B in the order of measurement. At this time, the control unit 13 also outputs the wearing article judgment result input from the operation unit 15 to the disease subject discrimination device 2B. The temperature data may be output each time the body surface temperature measuring device 1A measures a specific part, or may be output collectively after all specific parts of one subject have been measured.

[0084] The body surface temperature measuring device 1A may be provided with a liquid crystal display device or the like as a display unit that displays the measured temperature of the specific part. The liquid crystal display device may display the names of the specific parts to be measured in the order of measurement. For example, the user inputs that the temperature measurement of the specific part has ended by pressing the operation unit 15 every time the temperature of the specific part is measured using the body surface temperature measuring device 1A. When the operation unit 15 is pressed, the name of the specific part to be measured next may be displayed on the liquid crystal display device. Similarly, the liquid crystal display device may display information for inputting the determination result of the presence or absence of an article being worn. When the user, for example, operates the operation unit 15 to select and input the presence or absence of an article being worn, the operation unit 15 outputs the result of the article being worn to the control unit 13.

[0085] The diseased person discrimination device 2B can acquire temperature data of a specific body part from the body surface temperature measuring device 1A. In addition, the diseased person discrimination device 2B can transmit and receive data to and from the learning server 7 through the network N. The diseased person discrimination device 2B transmits to the learning server 7 the temperature data of the specific body part acquired by the body surface temperature measuring device 1A, the user discrimination result input by the user from the diseased person discrimination device 2B, and the worn article judgment result. In addition, the diseased person discrimination device 2B receives learned model data from the learning server 7.

[0086] The learning server 7 includes a communication unit 71, a temperature pattern learning unit 72, and a storage unit 73. The learning server 7 is capable of outputting to the diseased person discrimination device 2B a trained model trained using the temperature data of a specific part of the subject and the wearing garment determination result received from the diseased person discrimination device 2B, and the user discrimination result as input.

[0087] The communication unit 71 can be connected to the diseased person discrimination device 2B via the network N. The communication unit 71 receives temperature data of a specific part of the subject, a user discrimination result, and a worn article judgment result from the diseased person discrimination device 2B. The communication unit 71 also transmits a trained model read out from the storage unit 73 to the diseased person discrimination device 2B.

[0088] The temperature pattern learning unit 72 generates and updates a trained model from the learning result obtained by learning the temperature pattern based on the temperature data of the specific body part received by the communication unit 71 from the diseased person discrimination device 2B, and the user discrimination result and the worn clothing discrimination result input by the user. The learning process performed by the temperature pattern learning unit 72 is similar to the learning process performed by the temperature pattern learning unit 27 of the diseased person discrimination device 2 according to the first embodiment. Then, the temperature pattern learning unit 72 stores the updated trained model in the storage unit 73.

[0089] The memory unit 73 stores the trained model trained by the temperature pattern learning unit 72. As described above, the trained model read out from the memory unit 73 by the communication unit 71 is transmitted to the diseased person discrimination device 2B via the network N.

[0090] FIG. 8 is a block diagram showing an example of the internal configuration of the disease patient discrimination apparatus 2B. The diseased person discrimination device 2B has a configuration in which the specific part determination unit 21 and the temperature pattern learning unit 27 are removed from the diseased person discrimination device 2 shown in Fig. 3. The diseased person discrimination device 2B further includes a communication unit 28 capable of communicating with the learning server 7.

[0091] Based on the clothing judgment result, the temperature pattern calculation unit 22 calculates the temperature of the specific part measured in a specified measurement order by the user operating the body surface temperature measuring device 1A as a temperature pattern. The temperature pattern calculated by the temperature pattern calculation unit 22 is shaped for each person to be measured. The communication unit 28 transmits to the learning server 7 the temperature pattern calculated by the temperature pattern calculation unit 22, and the user identification result and the worn item determination result input by the user through the input unit 26. The communication unit 28 also receives the trained model from the learning server 7 and stores it in the memory unit 23.

[0092] The memory unit 23 stores the trained model and the worn article determination result received by the communication unit 28 from the learning server 7. The trained model and the worn article determination result are read out from the memory unit 23 by the subject discrimination unit 24 as appropriate and used to discriminate the subject.

[0093] The diseased person discrimination system 100A according to the third embodiment described above includes a body surface temperature measuring device 1A capable of measuring the temperature of each specific body part of the subject. A user brings one body surface temperature measuring device 1A close to a plurality of specific body parts of the subject in a specified measurement order, and obtains the temperature of each specific body part in turn. The user also inputs to the body surface temperature measuring device 1A whether or not the subject is wearing an article. The body surface temperature measuring device 1 outputs the temperature data and the article judgment result to the diseased person discrimination device 2B.

[0094] Furthermore, the learning server 7 provides the diseased person discrimination device 2B with an appropriate trained model obtained by machine learning the temperature pattern. Therefore, the diseased person discrimination device 2B can discriminate between a diseased person and a healthy person by selecting a trained model received from the learning server 7 according to the clothing judgment result and applying it to the temperature pattern. In this embodiment, the learning server 7 performs learning of the temperature pattern, which places a large load on the CPU 31 (see FIG. 4), so that the specifications of the diseased person discrimination device 2B can be lowered. Furthermore, since the body surface temperature measuring device 1A does not have a display unit 14, the body surface temperature measuring device 1A can be made smaller in size.

[0095] In this way, the diseased person discrimination device 2B is able to distinguish between diseased and healthy people without using thermal images as input, by inputting body surface temperature data measured for each specific body part from the body surface temperature measuring device 1A and the results of the clothing judgment.

[0096] Furthermore, the learning server 7 can learn temperature patterns that match the body surface temperatures of subjects measured in different regions by acquiring temperature patterns, clothing judgment results, and user discrimination results from multiple body surface temperature measuring devices 1A. This improves the accuracy of the temperature patterns, and also improves the accuracy of the discrimination results between sick and healthy subjects using these temperature patterns.

[0097] [Fourth embodiment] Next, a patient discrimination device used in a patient discrimination system according to a fourth embodiment of the present invention will be described with reference to Fig. 9. The patient discrimination device according to this embodiment determines, as a temperature pattern, the temperature of each pixel of a thermal image input from a body surface temperature measuring device that outputs a thermal image, and further discriminates a subject by applying a trained model selected based on a worn article determination result to the temperature pattern.

[0098] FIG. 9 is a block diagram showing an example of the internal configuration of a disease patient discrimination apparatus 2C. The diseased person discrimination device 2C has a configuration obtained by removing the specific part determination unit 21 from the diseased person discrimination device 2 shown in Fig. 3. In addition, the diseased person discrimination device 2C is connected to the body surface temperature measuring device 1 to constitute a diseased person discrimination system according to the fourth embodiment.

[0099] The temperature data input from the body surface temperature measuring device 1 to the temperature pattern calculation unit 22A includes the entire thermal image of the subject captured by the body surface temperature measuring device. In addition, the temperature pattern calculation unit 22A receives the worn article determination result from the body surface temperature measuring device 1. Like the specific part determination unit 21 according to the first embodiment, the temperature pattern calculation unit 22A has a function of aligning the facial features that make up the face based on the thermal image of the face. The worn article determination result is used for aligning the facial features.

[0100] Then, the temperature pattern calculation unit 22A calculates the temperature data extracted for each pixel of the thermal image by aligning the facial features that constitute the face using the clothing determination result based on the temperature data of the entire thermal image as a temperature pattern. The temperature pattern calculated by the temperature pattern calculation unit 22A is output to the subject discrimination unit 24 and the temperature pattern learning unit 27.

[0101] The operations of the memory unit 23, the subject discrimination unit 24, the display unit 25, the input unit 26 and the temperature pattern learning unit 27 are similar to the operations of each unit in the disease subject discrimination device 2 shown in Figure 3, so detailed explanations will be omitted.

[0102] In the diseased person discrimination system according to the fourth embodiment described above, the patient is discriminated as to whether the subject is a healthy person or a diseased person based on the temperature pattern calculated for each pixel of the thermal image input to the diseased person discrimination device 2C and the result of the wearing garment discrimination. Therefore, the process of determining a specific part of the subject from the thermal image is not required, and the process from obtaining the thermal image to discriminating the subject can be reduced.

[0103] [Fifth embodiment] Next, a disease discrimination device used in a disease discrimination system according to a fifth embodiment of the present invention will be described with reference to Figs. 10 and 11. In the disease discrimination system 100 according to the first embodiment described above, the presence or absence of an article is determined using only a thermal image of the subject, and a temperature pattern is calculated. However, in close proximity to each other, such as the subject's nose and inner corners of the eyes, the temperature difference is small, and the specific area may not be correctly identified. In addition, the edges of the article may be unclear in a thermal image, and the presence or absence of the article may not be correctly determined. Therefore, the body surface temperature measuring device according to this embodiment uses a visible image to determine the presence or absence of an article on the subject.

[0104] FIG. 10 is a block diagram showing an example of the internal configuration of the disease patient discrimination system 100B. The diseased person discrimination system 100B includes a body surface temperature measuring device 1B and a diseased person discrimination device 2.

[0105] Like the body surface temperature measuring device 1 shown in Figure 1, the body surface temperature measuring device 1B is equipped with a lens 10 capable of detecting infrared images, a detection element 11, an A / D conversion unit 12, a display unit 14, and an operation unit 15, and also includes a control unit 13A, a lens 40 capable of detecting visible images, a detection element 41, and an A / D conversion unit 42.

[0106] Lens 40 collects visible light emitted from the person being measured and forms an image on detection element 41 . The detection element 41 has a plurality of sensors 41a, and outputs a detection signal in a time-division manner for each frame period in synchronization with a vertical synchronization signal input to the detection element 41. The number of frames of the detection signal that the detection element 41 can output per unit time is, for example, 60 fps.

[0107] The sensor 41a of the detection element 41 is provided in pixel units. The sensor 41a detects the intensity of visible light imaged on the detection element 41 and outputs a detection signal. For example, a CCD imager or a CMOS sensor is used as the sensor 41a. The sensor 41a outputs an analog detection signal according to the intensity of the detected visible light.

[0108] The A / D conversion unit 42 converts the detection signal output from the pixel sensor 41a of the detection element 41 into digital data, and outputs the digital data to the control unit 13A.

[0109] The control unit 13A constructs a visible image from the digital data of the detection signal input from the A / D conversion unit 42. The control unit 13A then performs image processing such as edge enhancement on the visible image to obtain the contours of the face and body of the person being measured. If the person being measured is wearing glasses or a mask, the control unit 13A can obtain the contours of the article being worn and determine whether or not the article is being worn.

[0110] On the display unit 14, a thermal image showing the distribution state of the absolute temperature of the measurement object and a visible image are displayed superimposed or side-by-side based on the image data generated by the control unit 13A.

[0111] FIG. 11 is a block diagram showing an example of the internal configuration of the control unit 13A. The control unit 13A includes a gain correction unit 13a, an offset correction unit 13b, a calculation unit 13c, a thermal image generation unit 13d, an image processing unit 13f, and a worn article determination unit 13g.

[0112] The image processing unit 13f performs a predetermined image processing on the digital data of the visible image input from the A / D conversion unit 42. Examples of this image processing include AGC (Automatic Gain Control) and edge emphasis. The image processing unit 13f then outputs the visible image data after the image processing to the display unit 14 and the wearing article determination unit 13g. Therefore, the display unit 14 can display the visible image after the image processing.

[0113] The worn article determination unit 13g outputs a worn article determination result that determines whether or not the subject is wearing an article based on the visible image data (visible image) input from the A / D conversion unit 42. In this embodiment, the worn article of the subject includes, for example, at least one of a mask and glasses. When the worn article determination unit 13g determines that the subject is not wearing either a mask or glasses, it obtains a determination result that the subject is wearing an article. On the other hand, when the worn article determination unit 13g determines that the subject is wearing at least one of a mask and glasses, it obtains a determination result that the subject is wearing an article. The determination result that the subject is wearing an article or that the subject is wearing an article is output to the diseased person discrimination device 2 as a worn article determination result. The control unit 13A outputs the thermal image and the worn article determination result to the diseased person discrimination device 2. The control unit 13A may output the worn article determination result to the display unit 14, thereby displaying the worn article determination result on the display unit 14.

[0114] In the diseased person discrimination system 100B according to the fifth embodiment described above, the body surface temperature measuring device 1B automatically determines the presence or absence of an article worn by the subject using a visible image. This can improve the accuracy of the article determination result. The diseased person discrimination device 2 can apply an appropriate trained model to the temperature pattern calculated from the thermal image based on the article determination result determined using the visible image.

[0115] Furthermore, the body surface temperature measuring device 1B displays a visible image together with a thermal image on the display unit 14, allowing the user to check the garment judgment result of the subject by looking at the display unit 14. This allows the user to quickly respond when a problem occurs with the garment judgment result.

[0116] The configuration for capturing a visible image and obtaining a worn article determination result may be incorporated in the diseased person discrimination device 2A according to the second embodiment, the diseased person discrimination system 100A according to the third embodiment, and the diseased person discrimination device 2C according to the fourth embodiment. In each embodiment, it is also possible to apply an appropriate trained model to the temperature pattern calculated from the thermal image based on the worn article determination result determined using the visible image.

[0117] [Variations] Note that multiple non-contact thermometers (radiation thermometers) may be used as the body surface temperature measuring device. The multiple non-contact thermometers may be used to simultaneously measure multiple specific parts of the subject, thereby obtaining the temperatures of the specific parts for each subject.

[0118] Furthermore, the present invention is not limited to the above-described embodiments, and it goes without saying that various other applications and modifications are possible without departing from the gist of the present invention as set forth in the claims. For example, the above-mentioned embodiments have described the configurations of the devices and systems in detail and specifically in order to explain the present invention in an easily understandable manner, and are not necessarily limited to those having all of the configurations described. In addition, it is possible to replace part of the configurations of the embodiments described here with the configurations of other embodiments, and it is also possible to add the configurations of other embodiments to the configurations of one embodiment. In addition, it is also possible to add, delete, or replace part of the configurations of each embodiment with other configurations. In addition, the control lines and information lines shown are those that are considered necessary for the explanation, and not all control lines and information lines in the product are necessarily shown. In reality, it can be considered that almost all components are connected to each other. [Explanation of symbols]

[0119] 1...body surface temperature measuring device, 2...patient discrimination device, 13...control unit, 13e...wearing item determination unit, 14...display unit, 15...operation unit, 21...specific body part determination unit, 22...temperature pattern calculation unit, 23...storage unit, 24...measurement subject discrimination unit, 25...display unit, 26...input unit, 27...temperature pattern learning unit, 100...patient discrimination system

Claims

1. a temperature pattern calculation unit that calculates a temperature pattern of the temperature in the exposed part of the subject based on temperature data input from a body surface temperature measurement unit that measures the body surface temperature of the subject and a wearing article determination result that determines whether the subject is wearing an article, the temperature pattern being represented by a group of temperature data collected from the exposed part of the subject other than the part where the article is worn when the wearing article determination result is that an article is present, and the temperature pattern being represented by a group of temperature data collected from the exposed part of the subject when the wearing article determination result is that an article is not worn; and a subject discrimination unit that applies one trained model selected from a plurality of trained models that differ depending on whether the wearing subject determination result is a case where wearing subject is present or a case where wearing subject is absent to the temperature pattern based on the wearing subject determination result, thereby discriminating whether the subject is a sick person or a healthy person. Disease person identification device.

2. Further, a specific part determination unit is provided for determining a specific part of an exposed part of the subject based on the wearing article determination result and the temperature data, The temperature pattern calculation unit calculates the temperature pattern represented by a group of the temperature data collected from a plurality of the specific portions. The disease discrimination device according to claim 1 .

3. The temperature data includes a thermal image of an exposed part of the subject captured by the body surface temperature measuring unit, the specific part determination unit uses the wearing item determination result to align facial features constituting a face with respect to the thermal image to determine the specific part from the thermal image; The temperature pattern calculation unit extracts the temperature data from the determined specific portion and obtains the temperature patterns of a plurality of specific portions. The disease discrimination device according to claim 2 .

4. The temperature data includes a thermal image of an exposed part of the subject captured by the body surface temperature measuring unit, The temperature pattern calculation unit uses the wearing item determination result to align facial features constituting a face and calculates the temperature data extracted for each pixel of the thermal image as the temperature pattern. The disease discrimination device according to claim 1 .

5. The temperature data includes temperatures measured by the body surface temperature measuring unit from a specific part of the subject in a predetermined measurement order, The temperature pattern calculation unit calculates the temperatures of the specific parts measured in the measurement order as the temperature pattern based on the wearing article determination result. The disease discrimination device according to claim 1 .

6. Further, a storage unit that stores the trained model is provided, The subject discrimination unit discriminates the subject using the trained model read from the storage unit based on the wearing article discrimination result. The disease discrimination device according to any one of claims 1 to 5.

7. Further, the device includes a display unit, The subject discrimination unit displays, on the display unit, a discrimination result by the subject discrimination unit or a possibility that the subject is a patient. The disease discrimination device according to claim 6.

8. an input unit for inputting a user discrimination result indicating whether the subject is a healthy person or a diseased person; and a temperature pattern learning unit that generates a plurality of learned models trained by a machine learning model based on the user discrimination result, the wearing article determination result, and the temperature pattern used by the subject discrimination unit. The disease discrimination device according to claim 6.

9. The wearable article is at least one of a mask and glasses, The wearing item determination result is a result of determining whether the subject is wearing at least one of a mask, glasses, a collar, a scarf, and gloves as the wearing item. The disease discrimination device according to any one of claims 1 to 8.

10. The specific part includes at least one of the subject's eyes, the tip of the nose, the sides of the nose, the cheeks, the forehead, the temples, the neck, and the hands. The disease discrimination device according to claim 2 , 3 or 5 .

11. The body surface temperature measuring unit has a wearing article determination unit that determines the presence or absence of the wearing article based on the thermal image and outputs the wearing article determination result. The disease discrimination device according to claim 3 or 4.

12. The body surface temperature measuring unit has a wearing article determination unit that determines the presence or absence of the wearing article based on a visible image and outputs the wearing article determination result. The disease discrimination device according to any one of claims 1 to 10.

13. A disease discrimination system including a disease discrimination device and a learning server, The diseased person discrimination device comprises: a temperature pattern calculation unit that calculates a temperature pattern of the temperature in the exposed part of the subject based on temperature data input from a body surface temperature measurement unit that measures the body surface temperature of the subject and a wearing article determination result that determines whether the subject is wearing an article, the temperature pattern being represented by a group of temperature data collected from the exposed part of the subject other than the part where the article is worn when the wearing article determination result is that an article is present, and the temperature pattern being represented by a group of temperature data collected from the exposed part of the subject when the wearing article determination result is that an article is not worn; and a subject discrimination unit that applies one learned model selected from a plurality of learned models that differ depending on whether the wearing subject determination result is a case where wearing subject is present or a case where wearing subject is absent to the temperature pattern based on the wearing subject determination result, and discriminates whether the subject is a sick person or a healthy person, The learning server includes: A temperature pattern learning unit learns the temperature pattern according to the wearing article determination result by a machine learning model and generates a plurality of the learned models according to the wearing article determination result. Disease identification system.

14. The disease discrimination device has an input unit into which a user discrimination result indicating whether the subject is a healthy person or a diseased person is input, The temperature pattern learning unit generates the trained model based on the user identification result, the wearing article determination result, and the temperature pattern used by the measurement subject identification unit. The disease diagnosis system according to claim 13.

Images