Infrared temperature detection device, air conditioner and control method of air conditioner

Abstract

The invention discloses an infrared temperature detection device, an air conditioner and a control method of the air conditioner. The method comprises the steps that an infrared image of a to-be-detected user, the environment temperature, the equipment temperature of an infrared camera and the current distance of the to-be-detected user are obtained; calculating lumen value data and a background lumen value of the user to be measured according to the infrared image, calculating a maximum value of a human body lumen value and a lumen difference value of the background lumen value, and inputting the lumen value data, the lumen difference value, the environment temperature, the equipment temperature and the current distance into a blackbody lumen mapping model to obtain fitted blackbody lumen data; calculating a difference value between a black body lumen value and a background lumen value of different parts in the black body lumen data to obtain lumen difference value data; and inputting the lumen difference data, the blackbody lumen value, the environment temperature, the equipment temperature and the current distance into a human body surface temperature model for calculation to obtain human body temperature data output. According to the scheme, the influence of the object surface radiance on temperature detection can be reduced, and the temperature measurement precision is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of temperature detection, and particularly to an infrared temperature detection device, an air conditioner, and a control method for an air conditioner. Background Art

[0002] Compared with traditional temperature measurement technologies, infrared temperature measurement technology has the advantages of high sensitivity, fast speed, wide temperature measurement range, etc., and is thus widely used in fields such as home appliances and medical treatment. An infrared camera can obtain an infrared image of a measured target in a non-contact and long-distance manner, and can also measure the surface temperature of the target according to the output image.

[0003] However, in the actual use process, in the existing infrared human body temperature measurement system, due to the change in the surface emissivity of the detected object, the accuracy of the temperature output by the infrared camera will be affected, resulting in a low temperature measurement accuracy. Summary of the Invention

[0004] The purpose of the embodiments of the present invention is to provide an infrared temperature detection device, an air conditioner, and a control method for an air conditioner, which can reduce the influence of the surface emissivity of an object on temperature detection and improve the temperature measurement accuracy.

[0005] The embodiments of the present invention provide an infrared temperature detection device, including:

[0006] An infrared camera, configured to detect an infrared image of a user to be measured;

[0007] A first temperature sensor, configured to detect the current ambient temperature;

[0008] A second temperature sensor, configured to detect the device temperature of the infrared camera;

[0009] A distance sensor, configured to detect the current distance of the user to be measured;

[0010] A controller, configured to:

[0011] Obtain the infrared image, the ambient temperature, the device temperature, and the current distance;

[0012] Calculate the lumen value data and the background lumen value of the user to be measured according to the infrared image, and calculate the maximum value of the human lumen value and the lumen difference value of the background lumen value, where the lumen value data includes the lumen values of different parts of the user to be measured;

[0013] Input the lumen value data, the lumen difference value, the ambient temperature, the device temperature, and the current distance into a pre-trained blackbody lumen mapping model to obtain the fitted blackbody lumen data;

[0014] Calculate the difference between the blackbody lumen values at different parts in the blackbody lumen data and the background lumen value to obtain lumen difference data;

[0015] Input the lumen difference data, the blackbody lumen value, the ambient temperature, the device temperature, and the current distance into a pre-trained human surface temperature model for calculation to obtain an output of human temperature data.

[0016] Preferably, the controller is further configured to:

[0017] Pre-acquire a first training dataset and input the first training dataset into a pre-constructed neural network model for training;

[0018] Optimize the neural network model according to a pre-acquired second test dataset and a preset second objective function, search for model parameters corresponding to the optimal second objective function, and obtain the blackbody lumen mapping model;

[0019] Wherein, the first training dataset and the test dataset include lumen value data, ambient temperature, device temperature, current distance, and lumen difference as first model input data, and blackbody lumen data as first model output data.

[0020] As a preferred solution, the controller is further configured to:

[0021] Pre-acquire a second training dataset and input the second training dataset into a pre-constructed transfer learning model for training;

[0022] Optimize the neural network model according to a pre-acquired second test dataset and a preset second objective function, search for model parameters corresponding to the optimal second objective function, and obtain the human surface temperature model;

[0023] Wherein, the second training dataset and the test dataset include lumen difference data, blackbody lumen value, ambient temperature, device temperature, and current distance as second model input data, and human temperature data as second model output data.

[0024] Preferably, the process of acquiring the first training dataset or the first test dataset includes:

[0025] In a preset experimental environment, by changing the temperature of the blackbody, the distance between the blackbody and the infrared camera, and / or the current ambient temperature, acquire the current infrared image, and determine the current first model input data and the corresponding first model output data;

[0026] Use the current first model input data and the corresponding second model output data as the first training dataset or the first test dataset.

[0027] Preferably, the process of obtaining the second training data set or the second test data set includes:

[0028] In a preset experimental environment, by changing the distance between the experimenter and the infrared camera and / or the current environmental temperature, the current infrared image is obtained;

[0029] Determine the current second model input data, and read the current second model output data by setting temperature sensors at different parts of the experimenter;

[0030] Use the current second model input data and the corresponding second model output data as the second training data set or the second test data set.

[0031] As a preferred solution, the neural network model includes an input layer, two hidden layers, and an output layer;

[0032] The output of the neural network model

[0033] The first objective function is

[0034] Where A k (w, b, X) represents the calculation result of the k-th neuron in the hidden layer, p is the number of neurons in the hidden layer, σ(·) represents the Relu activation function, w is the weight, b is the bias, X is the set of input variables, x i represents the i-th input variable, w i represents the weight of the i-th input variable, q is the number of input variables; L1 is the first mean absolute error, S is the number of blackbody temperature samples in the test data set; z i is the predicted value calculated by the fully connected neural network model for the i-th model input data in the test data set, r i is the i-th model output data in the test data set.

[0035] Preferably, the transfer learning model includes a small neural network and a preset optimal model;

[0036] The small neural network includes an input layer, a hidden layer, and an output layer;

[0037] The output of the transfer learning model t = F(x1′, x2′, x′3, x′4, x′5);

[0038] The second objective function is

[0039] Where the i-th output of the small neural network i = 1, 2, …, 5, a k (w, b, X) represents the calculation result of the k-th neuron in the hidden layer, where w is the weight, b is the bias, X is the set of input variables, p is the number of neurons in the hidden layer, L2 is the second mean absolute error, and M is the number of human skin surface temperature samples in the second test dataset; y i is the predicted value calculated by the transfer learning model for the i-th model input data in the second test dataset, g i is the i-th model output data in the second test dataset.

[0040] Preferably, the background lumen value is specifically the average lumen value of several pixels at a preset pixel distance from the user to be measured in the infrared image;

[0041] The lumen value data includes the lumen values of the forehead, cheeks, and back of the hand of the user to be measured.

[0042] An embodiment of the present invention further provides an air conditioner, which includes:

[0043] The air conditioner body;

[0044] An infrared camera for detecting the infrared image of the user to be measured;

[0045] A temperature sensor for detecting the current ambient temperature;

[0046] A distance sensor for detecting the current distance of the user to be measured;

[0047] A controller configured to:

[0048] Obtain the infrared image, the ambient temperature, the device temperature, and the current distance;

[0049] Calculate the lumen value data and the background lumen value of the user to be measured according to the infrared image, and calculate the maximum value of the human lumen value and the lumen difference value of the background lumen value. The lumen value data includes the lumen values of different parts of the user to be measured;

[0050] Input the lumen value data, the lumen difference value, the ambient temperature, the device temperature, and the current distance into a pre-trained blackbody lumen mapping model to obtain the fitted blackbody lumen data;

[0051] Calculate the difference between the blackbody lumen values of different parts in the blackbody lumen data and the background lumen value to obtain lumen difference data;

[0052] Input the lumen difference data, the blackbody lumen value, the ambient temperature, the device temperature, and the current distance into a pre-trained human body surface temperature model for calculation to obtain an output of human body temperature data;

[0053] Match a corresponding control instruction in a preset control instruction matching library according to the human body surface temperature data, and control the operation of the air conditioner body according to the control instruction.

[0054] An embodiment of the present invention also provides a control method for an air conditioner, and the air conditioner includes:

[0055] An air conditioner body;

[0056] An infrared camera for detecting an infrared image of a user to be measured;

[0057] A temperature sensor for detecting the current ambient temperature;

[0058] A distance sensor for detecting the current distance of the user to be measured;

[0059] A controller;

[0060] The method includes:

[0061] Obtain the infrared image, the ambient temperature, the device temperature, and the current distance;

[0062] Calculate the lumen value data and the background lumen value of the user to be measured according to the infrared image, and calculate the maximum value of the human body lumen value and the lumen difference between the background lumen values. The lumen value data includes the lumen values of different parts of the user to be measured;

[0063] Input the lumen value data, the lumen difference, the ambient temperature, the device temperature, and the current distance into a pre-trained blackbody lumen mapping model to obtain fitted blackbody lumen data;

[0064] Calculate the difference between the blackbody lumen values of different parts in the blackbody lumen data and the background lumen value to obtain lumen difference data;

[0065] Input the lumen difference data, the blackbody lumen value, the ambient temperature, the device temperature, and the current distance into a pre-trained human body surface temperature model for calculation to obtain an output of human body temperature data;

[0066] Match a corresponding control instruction in a preset control instruction matching library according to the human body surface temperature data, and control the operation of the air conditioner body according to the control instruction.

[0067] Compared with the prior art, the infrared temperature detection device, air conditioner and control method of the air conditioner disclosed by the present invention include an infrared camera for detecting an infrared image of a user to be measured; a first temperature sensor for detecting the current ambient temperature; a second temperature sensor for detecting the device temperature of the infrared camera; a distance sensor for detecting the current distance of the user to be measured; and a controller configured to: obtain the infrared image, the ambient temperature, the device temperature and the current distance; calculate the lumen value data and the background lumen value of the user to be measured according to the infrared image, and calculate the maximum value of the human lumen value and the lumen difference value of the background lumen value, where the lumen value data includes the lumen values of different parts of the user to be measured; input the lumen value data, the lumen difference value, the ambient temperature, the device temperature and the current distance into a pre-trained blackbody lumen mapping model to obtain fitted blackbody lumen data; calculate the difference between the blackbody lumen values of different parts in the blackbody lumen data and the background lumen value to obtain lumen difference data; and input the lumen difference data, the blackbody lumen value, the ambient temperature, the device temperature and the current distance into a pre-trained human body surface temperature model for calculation to obtain an output of human body temperature data. The solution of the present application can reduce the influence of the object surface emissivity on temperature detection and improve the temperature measurement accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 is a schematic structural diagram of an infrared temperature detection device provided by an embodiment of the present invention;

[0069] Figure 2 is a schematic flowchart of the work executed by a controller provided by an embodiment of the present invention;

[0070] Figure 3 is another schematic flowchart of the work executed by a controller provided by an embodiment of the present invention;

[0071] Figure 4 is still another schematic flowchart of the work executed by a controller provided by an embodiment of the present invention;

[0072] Figure 5 Schematic structural diagram of a neural network model provided by an embodiment of the present invention

[0073] Figure 6 is a schematic structural diagram of a transfer learning model provided by an embodiment of the present invention;

[0074] Figure 7 is a schematic diagram of an infrared image provided by an embodiment of the present invention;

[0075] Figure 8 is a schematic structural diagram of an air conditioner body in an embodiment provided by an embodiment of the present invention;

[0076] Figure 9 is a partial structural schematic diagram of the refrigerant circuit of the air conditioner body in the embodiment of the present invention;

[0077] Figure 10 is another schematic flow chart of the work executed by the controller provided in the embodiment of the present invention. Detailed implementation manners

[0078] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts belong to the scope of protection of the present invention.

[0079] In the description of the present application, it should be understood that the terms "center", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present application and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation to the present application.

[0080] The terms "first" and "second" are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the present application, unless otherwise stated, the meaning of "plurality" is two or more.

[0081] In the description of the present application, it should be noted that unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense. For example, it may be a fixed connection, a detachable connection, or an integral connection; it may be a mechanical connection or an electrical connection; it may be directly connected, or indirectly connected through an intermediate medium, and it may be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to specific circumstances.

[0082] Refer to Figure 1 , which is a structural schematic diagram of the infrared temperature detection device provided in the embodiment of the present invention. The infrared detection device includes:

[0083] an infrared camera for detecting the infrared image of the user to be measured;

[0084] A first temperature sensor for detecting the current ambient temperature;

[0085] A second temperature sensor for detecting the device temperature of the infrared camera;

[0086] A distance sensor for detecting the current distance of the user to be measured;

[0087] A controller.

[0088] Specifically, the controller 140 is respectively connected to the infrared camera 150, the first temperature sensor 160, the second temperature sensor 170, and the distance sensor 180 to obtain data and perform corresponding control processes.

[0089] Among them, the infrared camera can obtain the infrared image of the target to be measured in a non-contact and long-distance situation, and can also measure the surface temperature of the target according to the output lumen value. Although infrared temperature measurement technology has many advantages, there are many factors that affect the accuracy of the lumen value output by the infrared camera. The main factors include stripe noise, ambient temperature, distance, etc. Stripe noise is formed when the sensors located in different columns on the infrared focal plane array use different readout circuits, and the difference in the bias voltage of the readout circuit generates light and dark on the infrared image. This strip noise usually appears as vertical stripes, and the stripes are relatively narrow, which will affect the maximum lumen value of the human body part in the infrared image, thereby affecting the temperature measurement accuracy. The ambient temperature will affect the internal ambient temperature of the system where the drive circuit board is located, and have a non-uniform effect on the output lumen value. The distance will directly affect the size of the target to be measured, make the generated image blurred, and also have a non-linear effect on the output lumen value, affecting the temperature measurement accuracy.

[0090] The solution of this application uses the first temperature sensor, the second temperature sensor, and the distance sensor to avoid these technical problems in the infrared temperature measurement device, and the user detects the influence of the ambient temperature, device temperature, and current distance on infrared temperature measurement.

[0091] Obtain the infrared image of the user to be measured through the infrared camera, and output the detected infrared image to the controller.

[0092] The first temperature sensor is deployed outside the housing of the infrared temperature detection device, and is used to detect the ambient temperature of the current environment and output the detected ambient temperature to the controller.

[0093] The second temperature sensor is deployed inside the infrared camera, and the user detects the device temperature inside the infrared camera and outputs the detected device temperature to the controller.

[0094] The distance sensor is close to the infrared camera device, and is used to detect the current distance between the user to be measured and the infrared camera and output the detected current distance to the controller.

[0095] The infrared temperature detection process is executed by the controller. During specific detection, refer to Figure 2 , which is a schematic flowchart of the work executed by the controller provided in an embodiment of the present invention. The controller executes the following steps:

[0096] Step S1, obtain the infrared image, the ambient temperature, the device temperature, and the current distance;

[0097] Step S2, calculate the lumen value data of the user to be measured and the background lumen value according to the infrared image, and calculate the maximum value of the human body lumen value and the lumen difference value of the background lumen value. The lumen value data includes the lumen values of different parts of the user to be measured;

[0098] Step S3, input the lumen value data, the lumen difference value, the ambient temperature, the device temperature, and the current distance into a pre-trained blackbody lumen mapping model to obtain the fitted blackbody lumen data;

[0099] Step S4, calculate the difference between the blackbody lumen values of different parts in the blackbody lumen data and the background lumen value to obtain lumen difference data;

[0100] Step S5, input the lumen difference data, the blackbody lumen value, the ambient temperature, the device temperature, and the current distance into a pre-trained human body surface temperature model for calculation to obtain the output of human body temperature data.

[0101] Specifically, refer to Figure 3 , which is another schematic flowchart of the work executed by the controller provided in an embodiment of the present invention. When the controller specifically performs temperature detection, it specifically executes the following steps;

[0102] Step S301, obtain the infrared image. That is, obtain the infrared image of the user to be measured through an infrared camera.

[0103] Step S302, calculate the lumen value data and the background lumen value in the infrared image, and calculate the maximum value of the human body lumen value and the lumen difference value of the background lumen value.

[0104] The maximum value of the human body lumen value is specifically the maximum lumen value in the human body area identified in the infrared image.

[0105] Due to different background temperatures, the background lumen values are also different, and the background environment changes complexly. It is difficult to fit the temperature simply using the background lumen value. It is found in the experiment that although the blackbody lumen values in three rounds of experiments under the same ambient temperature are different and the background lumen values are also different, the variable of the difference between the two is relatively stable. By adding this variable, the influence of the background temperature on the blackbody lumen value can be weakened, so that the neural network can more easily discover the change law, reduce the difference in the lumen value of the infrared image caused by the change in the emissivity of the detected object surface, and improve the detection accuracy.

[0106] Among them, the lumen value data includes the lumen values of different parts of the user to be measured. Specifically, when obtaining, according to the existing infrared image temperature measurement and recognition method, the human body in the infrared image can be recognized, different parts of the human body are marked, and the lumen values of different parts are obtained as the lumen value data. After recognizing the human body in the infrared image, the background lumen value can be obtained by identifying the pixels outside the area where the human body is located as the background area and determining the lumen value of the background area. By identifying the part interval of the user to be measured, that is, identifying a certain area as a certain part, the maximum lumen value in the area is used as the lumen value of the part, or the average value of the lumen values of all pixels in the area is used as the lumen value of the part, and the lumen values of different parts of the user to be measured are determined to obtain the lumen value data.

[0107] Among them, the background lumen value can also be obtained by identifying the background area outside the human body and taking the average value of the lumen values of the background area as the background lumen value.

[0108] Step S303, obtain the ambient temperature, device temperature, and current distance.

[0109] It should be noted that in this embodiment, the device temperature can specifically be the substrate temperature detected by the second temperature sensor configured on the circuit board of the infrared camera and / or the shutter temperature detected by the second temperature sensor configured on the shutter of the infrared camera. By using the substrate temperature and / or the shutter temperature as the device temperature, the influence of the infrared camera device temperature on the infrared image is considered, and the accuracy of temperature detection is improved.

[0110] Step S304, perform calculations using the blackbody lumen mapping model to obtain the fitted blackbody lumen data. That is, the lumen value data, lumen difference, ambient temperature, device temperature, and current distance are input into the pre-trained blackbody lumen mapping model to obtain the fitted blackbody lumen data.

[0111] The fitted blackbody lumen data is to map the lumen value data of the human body parts to the lumen values of the corresponding blackbodies through the blackbody lumen mapping model. The blackbody lumen data includes the blackbody lumen values of different parts.

[0112] Step S305: Calculate the lumen difference data, that is, calculate the difference between the blackbody lumen values at different parts of the blackbody lumen data and the background lumen value to obtain the lumen difference data at different parts.

[0113] Step S306: Calculate using the human body surface temperature model to obtain the output of human body temperature data. That is, input the lumen difference data, the blackbody lumen value, the ambient temperature, the device temperature, and the current distance into a pre-trained human body surface temperature model for calculation to obtain the output of human body temperature data.

[0114] When acquiring an infrared image, the radiation received by the infrared camera is not only the radiation of the target itself but also the reflected radiation of the target on the background environment. After being attenuated by the atmosphere, these radiations are received by the infrared camera. The unknown factors involved in the whole process are too complex. Therefore, the accuracy of target temperature detection is also affected by the radiation of the target on the background environment, resulting in the lumen value in the infrared image including the background lumen. Therefore, the lumen value is converted into the fitted blackbody lumen through the blackbody lumen mapping model, and the lumen value of the infrared image is standardized to the blackbody lumen to eliminate the influence of the background lumen and improve the temperature measurement accuracy.

[0115] When training the blackbody lumen mapping model, design a neural network to fit the blackbody lumen. Use the lumen value data, lumen difference, ambient temperature, device temperature, and current distance of the human body part in the infrared image as the input training data of the neural network, and use the output training data of the blackbody lumen obtained from the infrared image in the same environment for fitting training to obtain the fitted blackbody lumen mapping model.

[0116] When training the human body surface temperature model, design a neural network model to fit the skin surface temperature of the human body part. Use the lumen difference data, blackbody lumen value, ambient temperature, device temperature, and current distance in the infrared image as the input training data of the neural network, and use the skin surface temperature data of each part of the human body collected by the corresponding thermocouple as the output training data of the neural network for fitting training to obtain the fitted skin surface temperature detection model.

[0117] Use the powerful non-linear learning ability of the pre-trained skin temperature detection model to fit the non-uniformity law of the influence of ambient temperature, device temperature, and distance on the lumen value, so as to obtain the correct temperature of the human body part.

[0118] This application uses the maximum value of the lumen value and the background lumen value to eliminate the influence of the environmental background. Then, use the powerful non-linear learning ability of the neural network to fit the non-uniformity law of the influence of ambient temperature, device temperature, and distance on the lumen value, improve the temperature measurement accuracy, and thus obtain the correct temperature of the human body part.

[0119] It should be noted that in the embodiments of the present invention, the infrared image can also be filtered by using a Butterworth band - pass filter. That is, the infrared image is input into the Butterworth band - pass filter for filtering to obtain the filtered infrared image. The Butterworth band - pass filter is used to filter the stripe noise of the infrared image, remove the abnormal phenomenon of the lumen value caused by the noise, and improve the temperature measurement accuracy.

[0120] The controller is further configured to:

[0121] After obtaining the infrared image, perform a fast Fourier transform on the infrared image using a preset discrete Fourier transform formula, convert the infrared image from the spatial domain to the frequency domain space, and obtain an image spectrum;

[0122] Input the image spectrum into a preset Butterworth band - pass filter for filtering to obtain a filtered frequency - domain image;

[0123] Use the inverse discrete Fourier transform formula to transfer the frequency - domain image from the frequency domain space to the spatial domain to obtain a filtered infrared image.

[0124] When specifically performing temperature detection, the following steps are specifically executed;

[0125] When the controller filters the infrared image using a Butterworth band - pass filter, the following specific operations are performed:

[0126] Frequency - domain transformation: That is, perform a fast Fourier transform on the infrared image using a preset discrete Fourier transform formula, convert the infrared image from the spatial domain to the frequency domain space, and obtain an image spectrum. The principle of the Butterworth band - pass filter is based on the frequency - domain characteristics of the signal. The signal can be represented as a spectrum distribution, where components of different frequencies exhibit different amplitudes. The design goal of the Butterworth band - pass filter is to select the signal components within a specific frequency range and filter out the signal components of other frequencies. Therefore, the infrared image in the spatial domain needs to be transformed into the frequency domain first.

[0127] Butterworth band - pass filter filtering: Use the Butterworth band - pass filter to filter the spectrum of the image. Multiply the image spectrum in the frequency domain by the corresponding pixels of the Butterworth band - pass filter to obtain the filtered image in the frequency domain, that is, obtain a filtered frequency - domain image.

[0128] The performance of stripe noise in the spatial domain is relatively complex and it is not easy to find the pattern. However, in the frequency domain, this noise is a signal with a specific frequency. Therefore, by converting the image to the frequency domain and using the Butterworth band - pass filter to filter out this signal with a specific frequency, the purpose of denoising can be achieved.

[0129] Inverse transformation: After filtering the image spectrum in the frequency domain, it is necessary to transform the image in the frequency domain back to the spatial domain in order to read the corresponding lumen data. That is, the discrete Fourier inverse transform formula is used to transfer the frequency-domain image from the frequency domain space to the spatial domain, obtaining the filtered infrared image.

[0130] Filter the infrared image in the frequency domain through a Butterworth band-pass filter to eliminate the stripe noise caused by the bias voltage difference in the sensor array reading circuit of the infrared camera, and improve the accuracy of the infrared image.

[0131] Taking an infrared image as an example, first perform a fast Fourier transform on the image, and use a preset discrete Fourier transform formula to convert the image from the spatial domain to the frequency domain space;

[0132] The discrete Fourier transform formula is:

[0133]

[0134] where F(u, v) is the image spectrum, u and v are discrete frequency variables, u = 0, 1, 2, …, M - 1; v = 0, 1, 2, …, N - 1, f(x, y) is the lumen value in the spatial domain of the infrared image, which is a two-dimensional array, x and y are discrete real variables, representing the number of rows and columns corresponding to the two-dimensional array of the infrared image f(x, y) respectively, x = 0, 1, 2, …, M - 1; y = 0, 1, 2, …, N - 1, and M and N are the number of rows and columns corresponding to the two-dimensional array of the infrared image f(x, y) respectively.

[0135] Then, use a Butterworth band-pass filter to filter the spectrum of the image. The transfer function H(u, v) of the nth-order radially symmetric Butterworth band-pass filter is defined as:

[0136]

[0137] where W is the passband bandwidth, D0 is the passband center radius, where n is the order of the filter; D(u, v) is the band-pass filter in the circular domain with radius D0, D 2 (u, v) is the square of the filtering distance. Let the center point of this band-pass filter be point (u0, v0), D 2 (u, v) = (u - u0) 2 + (v - v0) 2 .

[0138] Multiply the data F(u, v) in the frequency domain by the corresponding pixels of the Butterworth band-pass filter H(u, v), and the filtered frequency-domain image G(u, v) = F(u, v)H(u, v) in the frequency domain can be obtained.

[0139] Use the inverse discrete Fourier transform formula to transfer the frequency-domain image from the frequency domain space to the spatial domain, and obtain the filtered infrared image.

[0140] The inverse discrete Fourier transform formula is:

[0141]

[0142] where g(x, y) is the filtered infrared image.

[0143] Complete the frequency-domain transformation, filtering, and spatial-domain transformation processes in sequence through the discrete Fourier transform formula, transfer function, and inverse discrete Fourier transform formula.

[0144] In another embodiment provided by the present invention, the controller is further configured to:

[0145] Pre-acquire a first training data set, and input the first training data set into a pre-constructed neural network model for training;

[0146] Optimize the neural network model according to a pre-acquired second test data set and a preset second objective function, find the model parameters corresponding to the optimal second objective function, and obtain the blackbody lumen mapping model;

[0147] wherein, the first training data set and the test data set include lumen value data, ambient temperature, device temperature, current distance, and lumen difference as the input data of the first model, and blackbody lumen values as the output data of the first model.

[0148] When specifically implementing this embodiment, refer to Figure 4 , which is another process schematic diagram of the work performed by the controller provided in the embodiment of the present invention. When the controller performs model training, the following steps are specifically executed:

[0149] Step S401, obtain a first training data set and a first test data set;

[0150] It should be noted that the first test data set and the first training data set are specifically obtained by dividing the same data set into different proportions to obtain the first test data set and the first training data set.

[0151] Step S402, input the first training data set into the neural network model for training. That is, input the lumen value data, ambient temperature, device temperature, current distance, and lumen difference as the input data of the model in the first training data set, and the blackbody lumen data as the output data of the first model into the neural network model for training respectively, and fit the relationship between the lumen value data, ambient temperature, device temperature, current distance, and lumen difference and the blackbody lumen data, where the blackbody lumen data includes the blackbody lumen values of different parts of the human body.

[0152] Step S403: Verify the current model using the first prediction dataset and a preset first objective function.

[0153] Step S404: Determine whether the verified first objective function is optimal.

[0154] Verify the maximum error of the model fitting the blackbody lumen value by predicting test data, determine whether the calculated error reaches a preset range, or whether the continuously calculated errors meet a preset optimal condition, so as to judge whether it reaches the optimal.

[0155] If not, return to step S401. During training, when the model does not reach the optimal, continue to obtain training data for training and optimize the model parameters until the model reaches the optimal.

[0156] If so, execute step S405;

[0157] Step S405: Obtain a blackbody lumen mapping model according to the model parameters of the current model.

[0158] Optimize the model through the objective function, output a blackbody lumen mapping model that meets the preset accuracy requirements, and use the output blackbody lumen mapping model for blackbody mapping, which can accurately eliminate the influence of background radiation.

[0159] In another embodiment provided by the present invention, the controller is further configured to:

[0160] Pre-obtain a second training dataset, and input the second training dataset into a pre-constructed transfer learning model for training;

[0161] Optimize the neural network model according to a pre-obtained second test dataset and a preset second objective function, find the model parameters corresponding to the optimal second objective function, and obtain the human body surface temperature model;

[0162] Wherein, the second training dataset and the test dataset include lumen difference data, blackbody lumen value, ambient temperature, device temperature, and current distance as input data of the second model, and human body temperature data as output data of the second model.

[0163] When specifically implementing this embodiment, when the controller performs model training, the following steps are specifically executed:

[0164] Obtain the second training dataset and the second test dataset;

[0165] It should be noted that the second test dataset and the second training dataset are specifically obtained by dividing the same dataset into different proportions to obtain the second test dataset and the second training dataset.

[0166] Input the second training dataset into the transfer learning model for training. That is, input the lumen difference data, blackbody lumen value, ambient temperature, device temperature, and current distance in the second training dataset as the model input data, and the human body temperature data as the second model output data into the transfer learning model for training respectively, to fit the relationship between the lumen difference data, blackbody lumen value, ambient temperature, device temperature, current distance, and human body temperature data, where the human body temperature data includes the temperatures of different parts of the human body.

[0167] Verify the current model using the second prediction dataset and a preset second objective function.

[0168] Determine whether the verified second objective function is optimal.

[0169] Verify the maximum error when the model fits the skin surface temperature of each part of the human body by predicting test data, determine whether the calculated error reaches the preset range, or whether the continuously calculated errors meet the preset optimal conditions, so as to judge whether it reaches the optimal.

[0170] If not, it indicates that the model has not reached the optimal. Continue to obtain training data for training and optimize the model parameters until the model reaches the optimal.

[0171] If so, obtain the human body surface temperature model according to the model parameters of the current model.

[0172] Optimize the model through the objective function, output the human body surface temperature model that meets the preset accuracy requirements, and use the output human body surface temperature model for temperature fitting to improve the accuracy of temperature detection.

[0173] In another embodiment provided by the present invention, the process of obtaining the first training dataset or the first test dataset includes:

[0174] In a preset experimental environment, by changing the temperature of the blackbody, the distance between the blackbody and the infrared camera, and / or the current ambient temperature, obtain the current infrared image, and determine the current first model input data and the corresponding first model output data;

[0175] Use the current first model input data and the corresponding second model output data as the first training dataset or the first test dataset.

[0176] In the specific implementation of this embodiment, this experiment is to obtain the lumen values of a blackbody at a fixed temperature and distance under different ambient temperatures. Considering the living rooms, bedrooms, studies commonly used by users and the temperature change ranges of various parts of the user's skin surface, the temperature change range of the blackbody is set from 25 degrees to 40 degrees, changing by one degree each time. The ambient temperature change range is set from 20 degrees to 32 degrees. Since it is difficult for the ambient temperature to reach a stable state, the ambient temperature changes by two degrees each time. The distance from the blackbody to the infrared camera lens is set from 100 cm to 350 cm, changing by 25 cm each time. Taking the experiment with an ambient temperature of 20 degrees as an example: Adjust the indoor ambient temperature to 20 degrees. After the indoor ambient temperature is maintained stable, set the blackbody temperature to 25 degrees, and obtain the infrared images of the blackbody at distances of 100 cm, 125 cm, 150 cm, up to 350 cm respectively. Subsequently, set the blackbody to 26 degrees and repeat the previous experiment until the blackbody temperature reaches 40 degrees to complete one round of experiments. Conduct three consecutive rounds of experiments at each ambient temperature, and try to ensure the stability of the ambient temperature in the three rounds of experiments. The purpose is to reduce the lumen value error caused by infrared noise in the experiment.

[0177] When the blackbody temperature is set to 25 degrees, the difference between the lumen values of the blackbody and the background lumen values in the three rounds of experiments is very large, approximately 70, but the variable of the difference between the blackbody lumen and the background lumen hardly differs in the three rounds of experiments.

[0178] Label the blackbody part, where the "heiti" label represents the blackbody part. By changing the temperature of the blackbody, the distance between the blackbody and the infrared camera, and / or the current ambient temperature, obtain the current infrared image, determine the current first model input data and the corresponding first model output data, and use the current first model input data and the corresponding second model output data as the first training dataset or the first test dataset.

[0179] In another embodiment provided by the present invention, the process of obtaining the second training dataset or the second test dataset includes:

[0180] In a preset experimental environment, by changing the distance between the experimenter and the infrared camera and / or the current ambient temperature, obtain the current infrared image;

[0181] Determine the current second model input data, and read the current second model output data by setting temperature sensors at different parts of the experimenter;

[0182] Use the current second model input data and the corresponding second model output data as the second training dataset or the second test dataset.

[0183] When this specific embodiment is implemented, when obtaining the second training dataset or the second test dataset for model training, a sealed space of about 10 - 15 square meters is built and a wall-mounted air conditioner is installed to create the environmental temperature changes in the bedroom in four seasons as the experimental environment.

[0184] Two experimenters sit on stools. The distance between the stools and the infrared camera is between 100 cm and 400 cm. The experimental environment simulates the heating and cooling processes when people use air conditioners indoors throughout the year. During the experiment, the indoor environmental temperature is increased and decreased uniformly. The environmental temperature at this time is read through the first temperature sensor, the current distance between the stool and the infrared camera is read through the distance sensor, the device temperature of the infrared camera is detected through the second temperature sensor, and the infrared image at this time is read through the infrared camera.

[0185] Thermocouples for collecting the surface temperature of the human body are pasted on the forehead, cheeks and back of the hands of the experimenters. The thermocouples record the temperatures of the three parts of the human body every second as the output human surface temperature data.

[0186] The current infrared image is input into a preset filter for filtering to obtain the filtered infrared image. The lumen values corresponding to the forehead, cheeks and back of the hands of the human body in the filtered infrared image are obtained as the lumen value data.

[0187] Through a preset number of experiments, in each experiment, one of the background lumen value, environmental temperature, device temperature and current distance is changed, and the temperatures of different parts are recorded as the output human surface temperature data to obtain the training output set.

[0188] In another embodiment provided by the present invention, the neural network model includes an input layer, two hidden layers and an output layer;

[0189] The output of the neural network model

[0190] The first objective function is

[0191] Among them, A k (w, b, X) represents the calculation result of the kth neuron in the hidden layer, p is the number of neurons in the hidden layer, σ(·) represents the Relu activation function, w is the weight, b is the bias, X is the set of input variables, x i represents the ith input variable, w i represents the weight of the ith input variable, q is the number of input variables; L1 is the first mean absolute error, S is the number of blackbody temperature samples in the test dataset; z iThe predicted value calculated by using the fully connected neural network model for the input data of the i-th model in the test dataset, r i is the output data of the i-th model in the test dataset.

[0192] When specifically implementing this embodiment, refer to Figure 5 , which is a schematic structural diagram of the neural network model provided by the embodiment of the present invention; the neural network model is composed of an input layer, two hidden layers, and an output layer.

[0193] This network has a total of 5 input variables, namely the lumen value data x1 of the human body part, the environmental temperature x2, the device temperature x3, the current distance x4, and the lumen difference x5. The five input variables are input into the neurons of the input layer, and after passing through two hidden layers and an output layer, the output result z is obtained.

[0194] Each hidden layer is composed of multiple neurons. The first hidden layer includes p neurons A 1 1 to A 1 p , and the second hidden layer includes p neurons A 2 1 to A 2 p .

[0195] The output of the fully connected neural network model

[0196] Among them, A k (w, b, X) represents the calculation result of the k-th neuron in the hidden layer. p is the number of neurons in the hidden layer. σ(·) represents the Relu activation function. w is the weight, b is the bias, X is the input variable, and x i represents the i-th input variable, and w i represents the weight of the i-th input variable. q is the number of input variables, which is 5 in this application.

[0197] The objective function of the mean absolute error used to optimize the model is

[0198] Among them, L1 is the mean absolute error, S is the number of skin surface temperature samples of each part in the test dataset; z i is the predicted value calculated by using the fully connected neural network model for the input data of the i-th model in the test dataset, r i is the output data of the i-th model in the test dataset.

[0199] In another embodiment provided by the present invention, the transfer learning model includes a small neural network and a preset optimal model;

[0200] The small neural network includes an input layer, a hidden layer, and an output layer;

[0201] The output t of the transfer learning model is t = F(x1′, x2′, x′3, x′4, x′5);

[0202] The second objective function is

[0203] where the i-th output of the small neural network i = 1, 2, …, 5, a k (w, b, X) represents the calculation result of the k-th neuron in the hidden layer, w is the weight, b is the bias, X is the set of input variables, p is the number of neurons in the hidden layer, L2 is the second mean absolute error, and M is the number of human skin surface temperature samples in the second test dataset; y i is the predicted value calculated by the transfer learning model for the i-th model input data in the second test dataset, and g i is the i-th model output data in the second test dataset.

[0204] In the specific implementation of this embodiment, refer to Figure 6 , which is the structural schematic diagram of the transfer learning model provided by the embodiment of the present invention.

[0205] The transfer model includes a first part and a second part. The first part is a small neural network, which consists of an input layer, a hidden layer, and an output layer. The n neurons A in the hidden layer 1 1 to A 1 n .

[0206] The second part is the optimal model F for fitting the blackbody temperature. The input of the optimal model F is obtained from the output of the small neural network, and the output is the skin surface temperature of the fitted human body part.

[0207] The i-th output of the small neural network i = 1, 2, …, 5, a k (w, b, X) represents the calculation result of the k-th neuron in the hidden layer, and x1 to x5 respectively represent the five input values of the small neural network.

[0208] The output t of the transfer learning model is t = F(x1′, x2′, x′3, x′4, x′5);

[0209] The second objective function is

[0210] L2 is the second mean absolute error, and M is the number of human skin surface temperature samples in the second test dataset; y iis the predicted value calculated by the transfer learning model for the i-th model input data in the second test dataset, g i is the i-th model output data in the second test dataset.

[0211] In another embodiment provided by the present invention, the background lumen value is specifically the average lumen value of several pixels at a preset pixel distance from the user to be measured in the infrared image;

[0212] The lumen value data includes the lumen values of the forehead, cheeks, and back of the hand of the user to be measured.

[0213] When specifically implementing this embodiment, refer to Figure 7 , which is a schematic diagram of an infrared image provided by an embodiment of the present invention.

[0214] The background lumen value is calculated from the average of 3 pixels each located above, below, left, and right outside the area where the identified user to be measured is located in the infrared image, that is Figure 7 In, outside the area where the user to be measured A is located, randomly select three pixel points a1, a2, and a3 from all pixel points that are 3 pixel points above the user to be measured and at the edge of the area where the user to be measured is located; randomly select three pixel points a4, a5, and a6 from all pixel points that are 3 pixel points below the user to be measured and at the edge of the area where the user to be measured is located; randomly select three pixel points a7, a8, and a9 from all pixel points that are 3 pixel points to the left of the user to be measured and at the edge of the area where the user to be measured is located; randomly select three pixel points a10, a11, and a12 from all pixel points that are 3 pixel points to the right of the user to be measured and at the edge of the area where the user to be measured is located. Calculate the average of the lumen values of pixel points a1 to a12 as the background lumen value.

[0215] It should be noted that in this embodiment, 3 pixels at a distance of 3 pixels are taken in the four directions of above, below, left, and right outside the area where the user to be measured is located. In other embodiments, M pixels can be randomly selected from all pixels at a distance of N pixels outside the area where the user to be measured is located, and the average value is calculated without distinguishing directions. The pixel distance and the number of pixel points can also be adjusted according to the actual situation.

[0216] It is also possible to determine how many pixels to take according to the distance. For example, select a point in the background of each part of the human body, and take infrared images of this point and the human body at different distances. On the infrared image, the distance between this point and each part of the human body changes. The farther the distance, the smaller the distance between this point and each part of the human body, and the background lumen value can be taken as the average of two pixels around each part of the human body; the closer the distance, the larger the distance between this point and each part of the human body, and the background lumen value can be taken as the average of six pixels around each part of the human body.

[0217] By taking the average value of the background pixels around the human body, the error of the background lumen value can be reduced.

[0218] When specifically performing temperature detection, by detecting the lumen values of the user's exposed forehead, cheeks, and the back of the hand, the temperature of different parts of the user can be determined to achieve the monitoring of the user's body temperature, facilitating the monitoring of the user's current state and being applicable in fields such as household appliances and medical treatment.

[0219] Another embodiment of the present invention provides an air conditioner, which includes:

[0220] An air conditioner body;

[0221] An infrared camera for detecting the infrared image of the user to be measured;

[0222] A first temperature sensor for detecting the current ambient temperature;

[0223] A second temperature sensor for detecting the device temperature of the infrared camera;

[0224] A distance sensor for detecting the current distance of the user to be measured;

[0225] A controller.

[0226] In the specific implementation of this embodiment, the air conditioner includes an air conditioner body, an infrared camera, a first temperature sensor, a second temperature sensor, a distance sensor, and a controller.

[0227] Refer to Figure 8 , which is a schematic structural diagram of an air conditioner body provided by an embodiment of the present invention in an implementation manner. An embodiment of the present invention provides an air conditioner body 100, which includes an indoor unit 110 and an outdoor unit 120. The indoor unit 110 is usually installed indoors and can be in the form of an indoor wall-mounted unit, an indoor cabinet unit, etc. The outdoor unit 120 is usually installed outdoors and is used for heat exchange in the indoor environment. The air conditioner 100 has a refrigerant circuit 130. By circulating the refrigerant in the refrigerant circuit 130, a vapor compression refrigeration cycle can be executed. A connecting pipe is used to connect the indoor unit 110 and the outdoor unit 120 to form a refrigerant circuit for the refrigerant to circulate.

[0228] Refer to Figure 9, which is a partial structural schematic diagram of the refrigerant circuit of the air conditioner body in the embodiments of the present invention. In this application, the air conditioner performs a refrigeration cycle by using a compressor 131, an indoor heat exchanger 132, a throttle valve 133, and an outdoor heat exchanger 134. The refrigeration cycle includes a series of processes involving compression, condensation, expansion, and evaporation, and supplies refrigerant to the conditioned and heat-exchanged air. Among them, the indoor heat exchanger 132 is usually arranged in the indoor unit 110, the compressor 131 and the outdoor heat exchanger 134 are usually arranged in the outdoor unit 120, the throttle valve 133 can be arranged in the indoor unit 110 or the outdoor unit 120, and the indoor heat exchanger 132 and the outdoor heat exchanger 134 are used as condensers or evaporators. When the indoor heat exchanger 132 is used as a condenser, the air conditioner serves as a heater in the heating mode, and when the indoor heat exchanger 132 is used as an evaporator, the air conditioner serves as a cooler in the cooling mode.

[0229] The compressor 131 compresses the refrigerant gas in a high-temperature and high-pressure state and discharges the compressed refrigerant gas. The discharged refrigerant gas flows into the condenser. The condenser condenses the compressed refrigerant into a liquid phase, and heat is released to the surrounding environment through the condensation process. The throttle valve 133 expands the high-temperature and high-pressure liquid-phase refrigerant condensed in the condenser into a low-pressure liquid-phase refrigerant. The evaporator evaporates the refrigerant expanded in the throttle valve 133 and returns the refrigerant gas in a low-temperature and low-pressure state to the compressor 131. The evaporator can achieve a refrigeration effect by using the latent heat of evaporation of the refrigerant to perform heat exchange with the material to be cooled. During the entire cycle, the air conditioner can adjust the temperature of the indoor space.

[0230] The air conditioner provided in the embodiments of the present invention further includes: an infrared camera 150, a first temperature sensor 160, a second temperature sensor 170, and a distance sensor 180 connected to the controller.

[0231] An infrared image of the user to be measured is obtained through the infrared camera, and the detected infrared image is output to the controller.

[0232] The first temperature sensor is deployed outside the housing of the infrared temperature detection device and is used to detect the ambient temperature of the current environment and output the detected ambient temperature to the controller.

[0233] The second temperature sensor is deployed inside the infrared camera to detect the device temperature inside the infrared camera and output the detected device temperature to the controller.

[0234] The distance sensor, close to the infrared camera device, is used to detect the current distance between the user to be measured and the infrared camera and output the detected current distance to the controller.

[0235] The controller performs temperature monitoring and controls the operation of the air conditioner body according to the monitoring results;

[0236] See Figure 10 , which is another schematic flow chart of the work performed by the controller provided in the embodiment of the present invention. The controller performs the following steps:

[0237] Step S1001, obtain the infrared image, the ambient temperature, the device temperature, and the current distance;

[0238] Step S1002, calculate the lumen value data of the user to be measured and the background lumen value according to the infrared image, and calculate the maximum value of the human body lumen value and the lumen difference value of the background lumen value. The lumen value data includes the lumen values of different parts of the user to be measured;

[0239] Step S1003, input the lumen value data, the lumen difference value, the ambient temperature, the device temperature, and the current distance into a pre-trained blackbody lumen mapping model to obtain the fitted blackbody lumen data;

[0240] Step S1004, calculate the difference between the blackbody lumen values of different parts in the blackbody lumen data and the background lumen value to obtain lumen difference data;

[0241] Step S1005, input the lumen difference data, the blackbody lumen value, the ambient temperature, the device temperature, and the current distance into a pre-trained human body surface temperature model for calculation to obtain an output of human body temperature data.

[0242] Step S1006, match a corresponding control instruction in a preset control instruction matching library according to the human body surface temperature data, and control the operation of the air conditioner body according to the control instruction.

[0243] Obtain an infrared image. That is, obtain an infrared image of the user to be measured through an infrared camera.

[0244] Calculate the lumen value data and the background lumen value in the infrared image, and calculate the maximum value of the human body lumen value and the lumen difference value of the background lumen value.

[0245] The maximum value of the human body lumen value is specifically the maximum lumen value in the human body area identified in the infrared image.

[0246] Due to different background temperatures, the background lumen values are also different, and the background environment changes complexly. It is difficult to fit the temperature simply using the background lumen value. It is found in the experiment that although the blackbody lumen values of three rounds of experiments under the same ambient temperature are different and the background lumen values are also different, the difference between the two is a relatively stable variable. By adding this variable, the influence of the background temperature on the blackbody lumen value can be weakened, so that the neural network can more easily discover the variation law, reduce the difference in the lumen value of the infrared image caused by the change in the surface emissivity of the detected object, and improve the detection accuracy.

[0247] Among them, the lumen value data includes the lumen values of different parts of the user to be measured. Specifically, when obtaining, according to the existing infrared image temperature measurement and recognition method, the human body in the infrared image can be recognized, different parts of the human body can be marked, and the lumen values of different parts are obtained as the lumen value data. After recognizing the human body in the infrared image, the pixels outside the area where the human body is located can be recognized as the background area, and the lumen value of the background area can be determined. Among them, by recognizing the part interval of the user to be measured, that is, recognizing a certain area as a certain part, the maximum lumen value in the area is used as the lumen value of the part, or the average value of the lumen values of all pixels in the area is used as the lumen value of the part, and the lumen values of different parts of the user to be measured are determined to obtain the lumen value data.

[0248] Among them, the background lumen value can also be the average value of the lumen values of the background area according to the recognized background area outside the human body.

[0249] Obtain the ambient temperature, device temperature, and current distance.

[0250] It should be noted that in this embodiment, the device temperature can specifically be the substrate temperature detected and obtained by the second temperature sensor configured on the circuit board of the infrared camera, and / or the shutter temperature detected and obtained by the second temperature sensor configured on the shutter of the infrared camera. By using the substrate temperature and / or the shutter temperature as the device temperature, the influence of the infrared camera device temperature on the infrared image is considered, and the accuracy of temperature detection is improved.

[0251] Use the blackbody lumen mapping model for calculation to obtain the fitted blackbody lumen data. That is, the lumen value data, lumen difference, ambient temperature, device temperature, and current distance are input into the pre-trained blackbody lumen mapping model to obtain the fitted blackbody lumen data.

[0252] The fitted blackbody lumen data is to map the lumen value data of the human body part to the lumen value of the corresponding blackbody through the blackbody lumen mapping model. The blackbody lumen data includes the blackbody lumen values of different parts.

[0253] Calculate the lumen difference data, that is, calculate the difference between the blackbody lumen values of different parts in the blackbody lumen data and the background lumen value to obtain the lumen difference data of different parts.

[0254] Perform calculations using the human body surface temperature model to obtain the output of human body temperature data. That is, input the lumen difference data, the blackbody lumen value, the ambient temperature, the device temperature, and the current distance into a pre-trained human body surface temperature model for calculation to obtain the output of human body temperature data.

[0255] When acquiring an infrared image, the radiation received by the infrared camera is not only the radiation of the target itself, but also the reflected radiation of the target to the background environment. After being attenuated by the atmosphere, these radiations are received by the infrared camera. The unknown factors involved in the whole process are too complex. Therefore, the accuracy of target temperature detection is also affected by the radiation of the target to the background environment, resulting in the lumen value in the infrared image including the background lumen. Therefore, the lumen value is converted into a fitted blackbody lumen through the blackbody lumen mapping model, and the lumen value of the infrared image is normalized to the blackbody lumen to eliminate the influence of the background lumen and improve the temperature measurement accuracy.

[0256] When training the blackbody lumen mapping model, design a neural network to fit the blackbody lumen. Use the lumen value data, lumen difference, ambient temperature, device temperature, and current distance of the human body part in the infrared image as the input training data of the neural network, and use the output training data of the blackbody lumen obtained from the infrared image placed in the same environment for fitting training to obtain a fitted blackbody lumen mapping model.

[0257] When training the human body surface temperature model, design a neural network model to fit the skin surface temperature of the human body part. Use the lumen difference data, blackbody lumen value, ambient temperature, device temperature, and current distance in the infrared image as the input training data of the neural network, and use the skin surface temperature data of each part of the human body collected by the corresponding thermocouple as the output training data of the neural network for fitting training to obtain a fitted skin surface temperature detection model.

[0258] Use the powerful non-linear learning ability of the pre-trained skin temperature detection model to fit the non-uniformity law of the influence of ambient temperature, device temperature, and distance on the lumen value, so as to obtain the correct temperature of the human body part.

[0259] This application uses the maximum value of the lumen value and the background lumen value to eliminate the influence of the environmental background. Then, use the powerful non-linear learning ability of the neural network to fit the non-uniformity law of the influence of ambient temperature, device temperature, and distance on the lumen value, improve the temperature measurement accuracy, and thus obtain the correct temperature of the human body part.

[0260] The temperature monitoring process of steps S1001 to S1005 in the solution of this application is the same as the specific implementation process in the embodiment of the above infrared temperature measurement device, and will not be elaborated here.

[0261] After obtaining the human body surface temperature data, that is, obtaining the temperatures of different parts of the human body, according to the temperatures of different parts of the human body, corresponding control instructions are matched in a preset control instruction matching library, and the control instruction matching library includes the corresponding relationships between different control instructions and human body surface temperature data.

[0262] By monitoring the skin surface temperatures of different parts of the user using the air conditioner, the current state of the user is judged. It is also possible to judge whether the human skin surface temperature rises or falls according to the difference between the skin surface temperature at the end of the set time period and the skin surface temperature at the start, and judge the current state of the user. Then, intelligent air conditioner control is performed through the control logic to avoid the user's body temperature being too high or too low, so as to improve the user experience.

[0263] An embodiment of the present invention also provides a control method for an air conditioner, which is applied to the air conditioner. The air conditioner includes:

[0264] An air conditioner body;

[0265] An infrared camera for detecting the infrared image of the user to be measured;

[0266] A first temperature sensor for detecting the current ambient temperature;

[0267] A second temperature sensor for detecting the device temperature of the infrared camera;

[0268] A distance sensor for detecting the current distance of the user to be measured;

[0269] A controller;

[0270] The method includes:

[0271] Obtaining the infrared image, the ambient temperature, the device temperature, and the current distance;

[0272] Calculating the lumen value data of the user to be measured and the background lumen value according to the infrared image, and calculating the maximum value of the human lumen value and the lumen difference of the background lumen value. The lumen value data includes the lumen values of different parts of the user to be measured;

[0273] Inputting the lumen value data, the lumen difference, the ambient temperature, the device temperature, and the current distance into a pre-trained blackbody lumen mapping model to obtain the fitted blackbody lumen data;

[0274] Calculate the difference between the blackbody lumen values at different parts of the blackbody lumen data and the background lumen value to obtain lumen difference data;

[0275] Input the lumen difference data, the blackbody lumen value, the ambient temperature, the device temperature, and the current distance into a pre-trained human surface temperature model for calculation to obtain an output of human temperature data;

[0276] Match a corresponding control instruction in a preset control instruction matching library according to the human surface temperature data, and control the operation of the air conditioner body according to the control instruction.

[0277] It should be noted that the control method of an air conditioner provided in an embodiment of the present invention is the same as all the process steps executed by a controller of an air conditioner in the above embodiment, and the working principles and beneficial effects of the two correspond one by one, so details are not described herein again.

[0278] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM), etc.

[0279] The above is the preferred implementation manner of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements are also regarded as the protection scope of the present invention.

Claims

1. An infrared temperature detection device, characterized in that, Including: An infrared camera for detecting an infrared image of a user to be measured; A first temperature sensor for detecting the current ambient temperature; A second temperature sensor for detecting the device temperature of the infrared camera; A distance sensor for detecting the current distance of the user to be measured; A controller configured to: Obtain the infrared image, the ambient temperature, the device temperature, and the current distance; Calculate the lumen value data and the background lumen value of the user to be measured based on the infrared image, and calculate the maximum value of the human lumen value and the lumen difference of the background lumen value, where the lumen value data includes the lumen values of different parts of the user to be measured; Input the lumen value data, the lumen difference, the ambient temperature, the device temperature, and the current distance into a pre-trained blackbody lumen mapping model to obtain the fitted blackbody lumen data; Calculate the difference between the blackbody lumen values of different parts in the blackbody lumen data and the background lumen value to obtain lumen difference data; Input the lumen difference data, the blackbody lumen value, the ambient temperature, the device temperature, and the current distance into a pre-trained human body surface temperature model for calculation to obtain the output of the human body temperature data.

2. The infrared temperature detection device according to claim 1, characterized in that, The controller is further configured to: Pre-obtain a first training data set and input the first training data set into a pre-constructed neural network model for training; Optimize the neural network model according to a pre-obtained second test data set and a preset second objective function, search for the model parameters corresponding to the optimal second objective function to obtain the blackbody lumen mapping model; Wherein, the first training data set and the test data set include the lumen value data, the ambient temperature, the device temperature, the current distance, and the lumen difference as the first model input data, and the blackbody lumen data as the first model output data.

3. The infrared temperature detection device according to claim 1, characterized in that, The controller is further configured to: Pre-obtain a second training data set and input the second training data set into a pre-constructed transfer learning model for training; Optimize the neural network model according to a pre-obtained second test data set and a preset second objective function, search for the model parameters corresponding to the optimal second objective function to obtain the human body surface temperature model; Wherein, the second training data set and the test data set include the lumen difference data, the blackbody lumen value, the ambient temperature, the device temperature, and the current distance as the second model input data, and the human body temperature data as the second model output data.

4. The infrared temperature detection device according to claim 2, characterized in that The process of obtaining the first training data set or the first test data set includes: In a preset experimental environment, by changing the temperature of the blackbody, the distance between the blackbody and the infrared camera, and / or the current ambient temperature, obtain the current infrared image, and determine the current first model input data and the corresponding first model output data; Use the current first model input data and the corresponding second model output data as the first training data set or the first test data set.

5. The infrared temperature detection device according to claim 3, characterized in that, The process of obtaining the second training data set or the second test data set includes: In a preset experimental environment, by changing the distance between the experimenter and the infrared camera and / or the current ambient temperature, the current infrared image is obtained; Determine the current second model input data, and read the current second model output data by setting temperature sensors at different parts of the experimenter; Use the current second model input data and the corresponding second model output data as the second training dataset or the second test dataset.

6. The infrared temperature detection device according to claim 2, wherein The neural network model includes an input layer, two hidden layers, and an output layer; Output of the neural network model The first objective function is Among them, A k (w, b, X) represents the calculation result of the k-th neuron in the hidden layer, p is the number of neurons in the hidden layer, σ(·) represents the Relu activation function, w is the weight, b is the bias, X is the set of input variables, and x i represents the i-th input variable, and w i represents the weight of the i-th input variable, and q is the number of input variables; L1 is the first mean absolute error, and S is the number of blackbody temperature samples in the test dataset; z i is the predicted value calculated by the fully connected neural network model for the i-th model input data in the test dataset, and r i is the i-th model output data in the test dataset.

7. The infrared temperature detection device according to claim 3, characterized in that The transfer learning model includes a small neural network and a preset optimal model; The small neural network includes an input layer, a hidden layer, and an output layer; The output t of the transfer learning model is t = F(x′1,x′2,x′3,x′4,x′5); The second objective function is Among them, the i-th output of the small neural network a k (w, b, X) represents the calculation result of the k-th neuron in the hidden layer, where w is the weight, b is the bias, X is the set of input variables, p is the number of neurons in the hidden layer, L2 is the second mean absolute error, and M is the number of human skin surface temperature samples in the second test dataset; y i is the predicted value calculated by the transfer learning model for the i-th model input data in the second test dataset, and g i is the i-th model output data in the second test dataset.

8. The infrared temperature detection device according to claim 1, wherein, The background lumen value is specifically the average lumen value of several pixels at a preset pixel distance from the user to be measured in the infrared image; The lumen value data includes the lumen values of the forehead, cheeks, and back of the hand of the user to be measured.

9. An air conditioner, characterized in that, The air conditioner includes: An air conditioner body; An infrared camera for detecting the infrared image of the user to be measured; A temperature sensor for detecting the current ambient temperature; A distance sensor for detecting the current distance of the user to be measured; A controller configured to: Obtain the infrared image, the ambient temperature, the device temperature, and the current distance; Calculate the lumen value data and the background lumen value of the user to be measured according to the infrared image, and calculate the maximum value of the human lumen value and the lumen difference between the background lumen value. The lumen value data includes the lumen values of different parts of the user to be measured; Input the lumen value data, the lumen difference, the ambient temperature, the device temperature, and the current distance into a pre-trained blackbody lumen mapping model to obtain the fitted blackbody lumen data; Calculate the difference between the blackbody lumen values of different parts in the blackbody lumen data and the background lumen value to obtain lumen difference data; Input the lumen difference data, the blackbody lumen value, the ambient temperature, the device temperature, and the current distance into a pre-trained human body surface temperature model for calculation to obtain the output of the human body temperature data; Match the corresponding control instruction in a preset control instruction matching library according to the human body surface temperature data, and control the operation of the air conditioner body according to the control instruction.

10. A control method for an air conditioner, characterized in that, The air conditioner includes: An air conditioner body; An infrared camera for detecting the infrared image of the user to be measured; A temperature sensor for detecting the current ambient temperature; A distance sensor for detecting the current distance of the user to be measured; A controller; The method includes: Obtain the infrared image, the ambient temperature, the device temperature, and the current distance; Calculate the lumen value data and the background lumen value of the user to be measured according to the infrared image, and calculate the maximum value of the human lumen value and the lumen difference between the background lumen value. The lumen value data includes the lumen values of different parts of the user to be measured; Input the lumen value data, the lumen difference, the ambient temperature, the device temperature, and the current distance into a pre-trained blackbody lumen mapping model to obtain the fitted blackbody lumen data; Calculate the difference between the blackbody lumen values of different parts in the blackbody lumen data and the background lumen value to obtain lumen difference data; Input the lumen difference data, the blackbody lumen value, the ambient temperature, the device temperature, and the current distance into a pre-trained human body surface temperature model for calculation to obtain the output of human body temperature data; Match the corresponding control instruction in a preset control instruction matching library according to the human body surface temperature data, and control the operation of the air conditioner body according to the control instruction.

Images