Temperature measurement method and device, computer equipment and storage medium

Abstract

The invention provides a temperature measurement method and device, computer equipment and a storage medium, and belongs to the field of temperature measurement, and the method comprises the steps: obtaining sample data; constructing a plurality of relational expressions between brightness values and temperature according to the plurality of spectral response logarithmic ratios and a two-color temperature measurement method formula; constructing a unit structure for mapping the brightness value to the predicted temperature; determining a physical constraint according to the relational expression between the plurality of brightness values and the temperature; constructing a neural network model for mapping the brightness value to the predicted temperature according to the unit structure and the physical constraint; the neural network model is trained through the sample data, the unknown parameters are determined, and then a temperature prediction model is obtained; acquiring a temperature field image of the target object; and inputting the temperature field image of the target object into the temperature prediction model to obtain the temperature of the target object. The neural network system constructed through physical constraint reduces the dependence of standardization processing on a real temperature value, and improves the convenience of temperature measurement.

Description

Technical Field

[0001] The present invention belongs to the field of temperature measurement, and in particular relates to a temperature measurement method, device, computer equipment and storage medium. Background Art

[0002] In the prior art, the temperature measurement through images mainly adopts the two-color temperature measurement method. However, the brightness values ​​of the RGB three channels of the color image captured by the CCD camera will vary due to the characteristics of the CCD camera itself and the aperture shutter combination. This leads to the need to first standardize the two-color temperature measurement formula when calculating the temperature using the two-color temperature measurement mechanism so that it can adapt to the brightness values ​​of the RGB three channels of the input image, thereby calculating a more accurate temperature value.

[0003] However, the normalization process is actually to calibrate an unknown quantity in the two-color temperature measurement formula, but this calibration process requires manual measurement of enough real temperature values, which is very difficult to measure in many industrial situations. For example, it is very difficult to measure the temperature value at a specified position in a large furnace, and the temperature can only be measured by inserting thermocouples at a few limited positions on the furnace surface.

[0004] In addition, the brightness values ​​of the three channels of RGB color images are usually between [0, 255]. When using a CCD camera to shoot a flame image of a temperature field, in order to make the calculated temperature value more accurate, the camera parameters are usually adjusted to make the brightness values ​​of the three channels of the image not equal to 0 and 255, so as to avoid distortion of the brightness values ​​of the three channels of the image. However, every time the camera parameters are adjusted, the formula of the two-color temperature measurement method needs to be re-normalized, which greatly hinders the real-time application of existing technologies in different occasions or occasions with a large temperature distribution span. Summary of the invention

[0005] In order to solve the problem in the prior art that image temperature measurement requires multiple normalization processes, which causes inconvenience in temperature measurement, the present invention provides a temperature measurement method, device, computer equipment and storage medium.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] First, a temperature measurement method is provided, the method comprising:

[0008] Obtain sample temperature field images and sample data corresponding to real temperatures;

[0009] Extract the three-primary color brightness values ​​of the sample temperature field image, characterize multiple spectral response logarithmic ratios respectively through the three-primary color brightness values ​​and multiple unknown parameters, and then construct multiple relationship formulas between brightness values ​​and temperatures according to the multiple spectral response logarithmic ratios and the two-color temperature measurement method formula; the spectral response logarithmic ratio is the logarithmic ratio of the spectral response intensity of one primary color channel relative to that of another primary color channel;

[0010] constructing an equation between the plurality of relationship equations of brightness values ​​and temperature, and using the equation as a physical constraint;

[0011] A neural network model is constructed by using the unknown parameters as network weights to map brightness values ​​to predicted temperatures; the neural network model is trained using the sample temperature field image and physical constraints to determine the unknown parameters, and then the unknown parameters are substituted into a relationship between brightness values ​​and temperature to obtain a temperature prediction model;

[0012] Acquire a temperature field image of a target object;

[0013] The temperature field image of the target object is input into the temperature prediction model to obtain the temperature of the target object.

[0014] Optionally, the plurality of spectral response logarithmic ratios include a red-blue spectral response logarithmic ratio and a red-green spectral response logarithmic ratio;

[0015] The red and blue spectral response logarithmic ratio is:

[0016] K rb =ln(K r / K b );

[0017] The red-green spectral response logarithmic ratio is:

[0018] K rg =ln(K r / K g );

[0019] Among them, K r is the red channel spectral response coefficient, K g is the green channel spectral response coefficient, K b is the blue channel spectral response coefficient.

[0020] Optionally, constructing multiple relationship equations between brightness values ​​and temperatures according to multiple spectral response logarithmic ratios and two-color temperature measurement formulas includes:

[0021] The two-color temperature measurement formula is:

[0022]

[0023] Among them, C 2represents the second radiation constant; λ r , g , b Respectively represent the three primary color wavelengths of the temperature field image; T rg and T rb are the temperature measurement results of the red-green two-color temperature measurement method and the red-blue two-color temperature measurement method; R e , G e , B e Represent the brightness values ​​of the R channel, G channel, and B channel respectively; ε(λ,T) is a dimensionless parameter representing the spectral emissivity;

[0024] The red-blue spectral response logarithmic ratio and the red-green spectral response logarithmic ratio are respectively characterized by the brightness values ​​of the three primary colors and a plurality of unknown parameters, and then the relationship between the plurality of brightness values ​​and the temperature is determined as follows:

[0025]

[0026] Among them, a rg’ 、b rg’ 、c rg’ 、a rb’ 、b rb’ and c rb’ is the unknown parameter.

[0027] Optionally, constructing a neural network model that maps brightness values ​​to predicted temperatures using the unknown parameters as network weights includes:

[0028] Corresponding to the relationship between brightness value and temperature, a neural network model is constructed to map brightness value to predicted temperature, and the single-channel expression of the neural network model is obtained as follows:

[0029]

[0030] Among them, g 1 , g 2 and g 3 Both represent activation functions, a rg 、b rg 、c rg 、a rb 、b rb and c rb represents the network weight to be solved corresponding to the unknown parameters, C rg and C rb represents a constant, T' rg and T' rb Respectively represent the output of a single channel:

[0031] Corresponding to the number of the unknown parameters, a multi-channel neural network model is constructed according to the single-channel expression.

[0032] Optionally, the neural network model is trained by the sample temperature field image and physical constraints to determine the unknown parameters, and then the unknown parameters are substituted into the relationship between the brightness value and the temperature to obtain the temperature prediction model, including:

[0033] Inputting the sample temperature field image into the neural network model to obtain a predicted temperature;

[0034] Determining a physical loss between a plurality of predicted temperatures according to the physical constraint, training the neural network model with the goal of minimizing the physical loss, and determining a value of the unknown parameter;

[0035] The value of the unknown parameter is substituted into the relationship between the brightness value and the temperature, and a weight coefficient is determined according to a plurality of sample data to obtain a temperature prediction model.

[0036] Optionally, determining a weight coefficient according to a plurality of sample data to obtain a temperature prediction model includes:

[0037] Substituting a plurality of sample data into a relational expression between brightness value and temperature for simultaneous calculation, and determining a weight coefficient of the relational expression between brightness value and temperature;

[0038] Then, the weight coefficient is substituted into the relationship between the brightness value and the temperature to determine the temperature prediction model.

[0039] Secondly, a temperature measuring device is also provided, the device comprising:

[0040] An acquisition module is used to acquire a sample temperature field image and sample data corresponding to the real temperature;

[0041] A construction module is used to extract the three-primary color brightness values ​​of the sample temperature field image, and to characterize a plurality of spectral response logarithmic ratios respectively through the three-primary color brightness values ​​and a plurality of unknown parameters, and then to construct a plurality of relationship equations between brightness values ​​and temperature according to the plurality of spectral response logarithmic ratios and a two-color temperature measurement method formula; the spectral response logarithmic ratio is the logarithmic ratio of the spectral response intensity of one primary color channel relative to another primary color channel; to construct equations between the relationship equations of the plurality of brightness values ​​and temperature, and to use the equations as physical constraints; to construct a neural network model that maps brightness values ​​to predicted temperatures using the unknown parameters as network weights; to train the neural network model through the sample temperature field image and the physical constraints, to determine the unknown parameters, and then to substitute the unknown parameters into the relationship equation between brightness values ​​and temperature to obtain a temperature prediction model;

[0042] The acquisition module is also used to acquire a temperature field image of the target object;

[0043] The prediction module is used to input the temperature field image of the target object into the temperature prediction model to obtain the temperature of the target object.

[0044] In addition, a computer-readable storage medium is provided, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the temperature measurement method is implemented.

[0045] In addition, a computer device is also provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the temperature measurement method is implemented when the processor executes the program.

[0046] The temperature measurement method provided by the present invention has the following beneficial effects:

[0047] Firstly, the correspondence between brightness value and temperature is constructed through multiple unknown parameters, which reduces the parameters that need to be calibrated for temperature measurement in the prior art. Secondly, a neural network system is constructed through physical constraints, and the unknown parameters are determined by the neural network system. The losses used in this process are all physical constraint losses, and no regression loss is involved. Therefore, the dependence of the temperature prediction model on sample data is greatly weakened. This is not only conducive to improving the accuracy of temperature measurement, but also reduces the dependence of normalization processing on calibration parameters, improves the convenience of normalization processing, eliminates the obstacles to the real-time application of image temperature measurement in different occasions or occasions with a large temperature distribution span, and improves the convenience of temperature measurement. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiment of the present invention and its design scheme, the following briefly introduces the drawings required for this embodiment. The drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0049] Figure 1 The figure is a flow chart of a temperature measurement method provided by the present invention according to an exemplary embodiment.

[0050] Figure 2 A schematic diagram of a unit structure for mapping a brightness value to a predicted temperature according to an exemplary embodiment of the present invention;

[0051] Figure 3 A schematic diagram of a neural network model based on physical constraints provided according to an exemplary embodiment of the present invention.

[0052] Figure 4 The present invention is a block diagram of a temperature measurement device according to an exemplary embodiment. DETAILED DESCRIPTION

[0053] In order to enable those skilled in the art to better understand the technical solution of the present invention and implement it, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the scope of protection of the present invention.

[0054] There are three main radiation temperature measurement methods: brightness temperature measurement, full radiation temperature measurement and two-color temperature measurement. With the rapid development of CCD cameras and electronic computer technology, an image-based temperature measurement method has been born. This method collects the temperature field combustion flame image and uses the radiation temperature measurement mechanism to measure the temperature field temperature distribution.

[0055] The image-based radiation temperature measurement method has many advantages: (1) It has low thermal inertia and can realize transient temperature measurement; (2) This method is a non-contact temperature measurement method, which can avoid direct contact between the equipment and the harsh environment and protect the equipment from corrosion; (3) It can measure the temperature distribution of large temperature fields in real time and realize the visualization of the measured temperature field.

[0056] By adding a monochrome filter to the CCD camera, it is possible to quickly collect relatively accurate brightness values ​​of the RGB three channels of the color image. The two-color temperature measurement method can take advantage of this feature, combining the principle of the two-color temperature measurement method and the brightness values ​​of the RGB three channels of the color image to measure the temperature of the area photographed by the CCD camera, thereby realizing the visualization of the temperature distribution.

[0057] The present invention adds physical constraints on the basis of the two-color temperature measurement method, thereby improving the convenience of image temperature measurement.

[0058] The technical solutions provided by various embodiments of the present invention are described in detail below in conjunction with the accompanying drawings.

[0059] First, the present invention provides a temperature measurement method, specifically as follows Figure 1 As shown, the following steps are included:

[0060] S101, obtaining a sample temperature field image and sample data of its corresponding real temperature.

[0061] Specifically, a CCD camera may be used to capture a temperature field image, and denoising may be performed on the captured image, and the real temperature corresponding to the temperature field image may be measured.

[0062] S102. Construct multiple relationship equations between brightness values ​​and temperatures according to multiple spectral response logarithmic ratios and two-color temperature measurement method formulas.

[0063] Specifically, the three-primary-color brightness values ​​of the sample temperature field image can be extracted, and a variety of spectral response logarithmic ratios can be respectively characterized by the three-primary-color brightness values ​​and a plurality of unknown parameters. Then, a plurality of relationship equations between brightness values ​​and temperatures can be constructed according to a plurality of spectral response logarithmic ratios and the two-color temperature measurement method formula. The spectral response logarithmic ratio is the logarithmic ratio of the spectral response intensity of one primary color channel relative to that of another primary color channel.

[0064] In one embodiment, the plurality of spectral response logarithmic ratios include a red-blue spectral response logarithmic ratio and a red-green spectral response logarithmic ratio.

[0065] The red-to-blue spectral response log ratio is determined by the following formula:

[0066] K rb =ln(K r / K b );

[0067] The red-to-green spectral response log ratio is determined by the following formula:

[0068] K rg =ln(K r / K g );

[0069] Among them, K r is the red channel spectral response coefficient, K g is the green channel spectral response coefficient, K b is the blue channel spectral response coefficient. The logarithmic ratio of the spectral response is affected by the spectral response characteristics of the CCD and can be regarded as an error correction coefficient. It is a nonlinear function of the temperature value and is also the quantity that needs to be calibrated in the two-color temperature measurement formula.

[0070] In another embodiment, the temperature measurement formula for double-color hair is determined as follows:

[0071]

[0072] Among them, C 2 represents the second radiation constant, C 2 =1.4338×10-2m·K; λ r =700.0nm,λ g =546.1nm, λ b =435.8nm represents the wavelength of the three colors of RGB of the color CCD; T rg 、T rb and T gb They are the temperature measurement results of the radiator using the two-color method. When measuring the temperature of the same object, ideally, T rg =T rb =T gb, the unit is K; Re, Ge, Be represent the pixel values ​​of R channel, G channel and B channel respectively, also called brightness value; ε(λ,T) is a dimensionless parameter, which represents the spectral emissivity.

[0073] GB two-color temperature measurement will distort the temperature measurement result due to the saturation of the photogenerated charge in the R base color potential well, so the present invention selects RG and RB two-color temperature measurement methods for research. Since the flame is mainly composed of solid particles with continuous radiation spectrum, when measuring the temperature distribution of the combustion flame, the flame can be approximately treated as a gray body, that is, ln(ε(λ r ,T) / ε(λ g ,T))=0,ln(ε(λ r ,T) / ε(λ b ,T))=0. Since K rg and K rb There is a quadratic function relationship between the natural logarithm of the corresponding brightness ratio, and the red-blue spectral response logarithm ratio and the red-green spectral response logarithm ratio can be represented by the brightness values ​​of the three primary colors and multiple unknown parameters, and then the relationship between the multiple brightness values ​​and the temperature is determined as follows:

[0074]

[0075] Among them, a rg’ 、b rg’ 、c rg’ 、a rb’ 、b rb’ and c rb’ is the unknown parameter.

[0076] S103: construct an equation between the multiple relationship equations of brightness values ​​and temperature, and use the equation as a physical constraint.

[0077] In this step, considering that the existing technology of measuring temperature through images requires multiple normalization processes and uses a large amount of sample data to train the model, which makes the preliminary processing very cumbersome when applied across occasions, the present invention proposes the concept of physical constraints based on the above problems, constructs equations between the relationship between multiple brightness values ​​mapped to temperatures, constrains the temperatures predicted by the brightness values ​​of different primary colors, and then constructs a self-constraint model that does not require actual temperature measurement.

[0078] Specifically, first of all, considering that the pixel values ​​of the three color channels of the same pixel can only be integers in [0,255], there will definitely be slight errors. The relationship between the above brightness value and temperature can be converted into the following formula to improve the accuracy:

[0079]

[0080] Where Δb rgand Δb rb is the error, which is a very small amount, close to 0, and η rg and η rb is the error coefficient, which is approximately equal to 1.

[0081] Secondly, the temperature equations of the red-blue two-color temperature measurement method and the red-green two-color temperature measurement method are constructed; the mapping relationship of the unit structure is substituted into the temperature equation to obtain the physical constraint.

[0082] Specifically, multiple equations for the relationship between brightness values ​​and temperature are constructed as follows:

[0083]

[0084] Due to Δb rg and Δb rb is a very small quantity approaching 0, and η rg and η rb is approximately equal to 1, so the above equation can be simplified to:

[0085]

[0086] Substituting into the unit structure, we get the physical constraint T r ' g ≈T r ' b , the physical constraint is a constraint equation constructed by different two-color temperature measurement formulas.

[0087] S104, constructing a neural network model that maps brightness values ​​to predicted temperatures using the unknown parameters as network weights.

[0088] Specifically, a neural network model that maps brightness values ​​to predicted temperatures may be constructed corresponding to the relationship between brightness values ​​and temperature.

[0089] First, we can construct a basic unit structure that maps brightness values ​​to predicted temperatures. The basic unit structure is as follows: Figure 2 shown.

[0090] The basic unit structure is used as a single channel of the neural network model, corresponding to the number of unknown parameters, and a multi-channel neural network model is constructed according to the single-channel expression.

[0091] Among them, the single channel expression is:

[0092]

[0093] Among them, g 1 , g 2 and g 3 Both represent activation functions, g 1 (·)=ln(·),g 2(·)=(·) 2 , a rg 、b rg 、c rg 、a rb 、b rb and c rb Represents the network weight to be solved, respectively with a rg’ 、b rg’ 、c rg’ 、a rb’ 、b rb’ and c rb’ One-to-one correspondence, C rg and C rb represents a constant, T' rg and T' rb Respectively represent the output of the unit structure.

[0094] For example, when the RGB three-channel brightness value of a pixel is known, a pixel can construct a single-channel expression as above. It is known that the single-channel expression has 6 unknowns. In theory, 6 different equations need to be constructed by 6 pixels at different temperatures to solve the 6 unknowns. A unit structure can input one pixel at a time, so using 6 parallel unit structures at the same time can construct a neural network model based on physical constraints. Figure 3 shown.

[0095] Since the temperature values ​​measured by the red-blue two-color temperature measurement method and the red-green two-color temperature measurement method are theoretically equal, the present invention constructs a sub-model for the red-blue two-color temperature measurement method and the red-green two-color temperature measurement method according to the mechanism formula. The two sub-models will theoretically output temperature values ​​after inputting brightness values, and the temperature values ​​are equal in theory, but the two sub-models have three unknown quantities to be trained when they are constructed. Before training, the temperature values ​​output by the two sub-models are not equal, so these two different temperature values ​​will generate a physical constraint loss. Using the physical constraint as the loss function, a six-channel neural network model that maps brightness values ​​to predicted temperatures is constructed.

[0096] S105: training the neural network model to obtain a temperature prediction model.

[0097] Specifically, the neural network model is trained through the sample temperature field image and physical constraints to determine the unknown parameters, and then the unknown parameters are substituted into the relationship between the brightness value and the temperature to obtain the temperature prediction model.

[0098] In this step, the sample temperature field image in the sample data can be input into the neural network model to obtain the predicted temperature; the physical loss between multiple predicted temperatures is determined according to the physical constraint, and the neural network model is trained with the goal of minimizing the physical loss to determine the value of the unknown parameter; the value of the unknown parameter is substituted into the relationship between the brightness value and the temperature, and the weight coefficient is determined according to multiple sample data to obtain a temperature prediction model.

[0099] Among them, the input is a known quantity. At the same time, six pixels with different temperatures are input into the neural network model. After multiple moments, the model will gradually adjust the network weights due to the loss of physical constraints, that is, adjust the six unknown parameters, so that the physical loss of the model gradually approaches 0, and finally obtain a relatively accurate solution to the six unknown parameters.

[0100] Since the losses used in the training model are all physical constraint losses and do not involve regression losses, this method can greatly reduce the network model's dependence on the true value. In addition, because solving the six parameters can only satisfy the equation of the relationship between brightness value and temperature, simply solving the six parameters in the model does not have the ability to measure temperature. That is, multiplying both sides of the equation by the unknown coefficient at the same time, the equation still holds, as shown in the following formula:

[0101]

[0102] Among them, β is the unknown coefficient. By splitting the above equation, we can get:

[0103]

[0104] Among them, β rg and β rg are the weight coefficients of the relationship between different brightness values ​​and temperatures constructed based on the red-green two-color temperature measurement method and the red-blue two-color temperature measurement method, respectively. rg and α rb are the errors of the relationship between different brightness values ​​and temperatures.

[0105] Since the physical constraint is a constraint constructed by the equation of the relationship between different brightness values ​​and temperatures, that is, the physical loss it produces is not the loss between the predicted result and the true value, but the loss between the predicted results of the mapping relationship between different brightness values ​​and temperatures. This causes the network weights trained by the neural network model to be possibly not equal to the true network weights, but rather a related value that has a multiple or other relationship with the true network weights.

[0106] In order to further determine the true network weight, that is, the true value of the unknown parameter, multiple sample data can be substituted into the relationship between brightness value and temperature to determine the weight coefficient of the relationship between brightness value and temperature; and then the weight coefficient is substituted into the relationship between brightness value and temperature to determine the temperature prediction model.

[0107] For example, after substituting the value of the unknown parameter into the temperature equation, multiple actually measured sample data are substituted into the temperature equation for combination to determine the weight coefficient and error amount of the relationship between the brightness value and the temperature; the weight coefficient and error amount are substituted into the relationship between the brightness value and the temperature to determine the temperature prediction model.

[0108] Therefore, the weight coefficient can be obtained by substituting a plurality of sample data into the relationship between the brightness value and the temperature and performing a joint calculation.

[0109] Finally, the calculated weight coefficient is substituted into the relationship between the brightness value and the temperature to obtain the temperature prediction model.

[0110] S106. Obtain the temperature of the target object through a temperature prediction model.

[0111] Specifically, a temperature field image of the target object may be acquired; the temperature field image of the target object is input into the temperature prediction model to obtain the temperature of the target object.

[0112] In addition, in order to further improve the accuracy of the temperature prediction model, T rg and T rb The average value is taken as the final measurement result.

[0113] By adopting the above method, firstly, the correspondence between brightness value and temperature is constructed through multiple unknown parameters, which reduces the parameters that need to be calibrated for temperature measurement in the prior art. Secondly, a neural network system is constructed through physical constraints, and the unknown parameters are determined through the neural network system. The losses used in this process are all physical constraint losses, and no regression loss is involved. Therefore, the dependence of the temperature prediction model on sample data is greatly weakened. This is not only conducive to improving the accuracy of temperature measurement, but also reduces the dependence of normalization processing on calibration parameters, improves the convenience of normalization processing, eliminates the obstacles to the real-time application of image temperature measurement in different occasions or occasions with a large temperature distribution span, and improves the convenience of temperature measurement.

[0114] Secondly, the present invention also provides a temperature measuring device, such as Figure 2 As shown, including:

[0115] The acquisition module 201 is used to acquire the sample temperature field image and the sample data of the corresponding real temperature.

[0116] Construction module 202 is used to extract the three-primary color brightness values ​​of the sample temperature field image, and to characterize multiple spectral response logarithmic ratios respectively through the three-primary color brightness values ​​and multiple unknown parameters, and then to construct multiple relationship equations between brightness values ​​and temperature according to the multiple spectral response logarithmic ratios and the two-color temperature measurement method formula; the spectral response logarithmic ratio is the logarithmic ratio of the spectral response intensity of one primary color channel relative to another primary color channel; to construct a unit structure for mapping brightness values ​​to predicted temperatures; to determine a mapping relationship equation for the unit structure according to the multiple relationship equations between brightness values ​​and temperatures, and to determine physical constraints according to the mapping relationship equation; the physical constraints are constraint equations constructed through different two-color temperature measurement method formulas; to construct a neural network model for mapping brightness values ​​to predicted temperatures according to the unit structure and the physical constraints; to train the neural network model through the sample data, to determine the unknown parameters, and then to obtain a temperature prediction model.

[0117] The acquisition module 201 is also used to acquire a temperature field image of the target object.

[0118] The prediction module 203 is used to input the temperature field image of the target object into the temperature prediction model to obtain the temperature of the target object.

[0119] By adopting the above device, firstly, the correspondence between brightness value and temperature is constructed through multiple unknown parameters, thereby reducing the parameters that need to be calibrated for temperature measurement in the prior art. Secondly, a neural network system is constructed through physical constraints, and the unknown parameters are determined through the neural network system. The losses used in this process are all physical constraint losses, and regression losses are not involved. Therefore, the dependence of the temperature prediction model on sample data is greatly weakened. This is not only conducive to improving the accuracy of temperature measurement, but also reduces the dependence of normalization processing on calibration parameters, improves the convenience of normalization processing, eliminates obstacles to the real-time application of image temperature measurement in different occasions or occasions with a large temperature distribution span, and improves the convenience of temperature measurement.

[0120] The present invention also provides a computer-readable storage medium, which stores a computer program, which can be used to execute the above Figure 1 The steps of the temperature measurement method are provided.

[0121] The present invention also provides a computer device. At the hardware level, the computer device includes a processor, an internal bus, a network interface, a memory, and a non-volatile memory. Of course, it may also include hardware required for other services. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it to achieve the above Figure 1 The steps of the temperature measurement method are provided.

[0122] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0123] The present invention is described with reference to flowcharts and / or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as a combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0124] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0125] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0126] It should be noted that the above specific implementation method can enable those skilled in the art to understand the invention more comprehensively, but does not limit the invention in any way. Therefore, although the present invention has been described in detail in this specification, those skilled in the art should understand that the invention can still be modified or replaced by equivalents; and all technical solutions and improvements that do not deviate from the spirit and scope of the invention are included in the protection scope of the patent for the invention. Any figure mark in the claims should not be regarded as limiting the claims involved.

Claims

1. A temperature measurement method, characterized in that: The method comprises: Obtain sample temperature field images and sample data corresponding to real temperatures; Extract the three-primary color brightness values ​​of the sample temperature field image, characterize multiple spectral response logarithmic ratios respectively through the three-primary color brightness values ​​and multiple unknown parameters, and then construct multiple relationship formulas between brightness values ​​and temperatures according to the multiple spectral response logarithmic ratios and the two-color temperature measurement method formula; the spectral response logarithmic ratio is the logarithmic ratio of the spectral response intensity of one primary color channel relative to that of another primary color channel; constructing an equation between the plurality of relationship equations of brightness values ​​and temperature, and using the equation as a physical constraint; Constructing a neural network model that maps brightness values ​​to predicted temperatures using the unknown parameters as network weights; The neural network model is trained by using the sample temperature field image and physical constraints to determine the unknown parameters, and then the unknown parameters are substituted into the relationship between the brightness value and the temperature to obtain a temperature prediction model; Acquire a temperature field image of a target object; The temperature field image of the target object is input into the temperature prediction model to obtain the temperature of the target object.

2. A temperature measurement method according to claim 1, characterized in that: The multiple spectral response logarithmic ratios include a red-blue spectral response logarithmic ratio and a red-green spectral response logarithmic ratio; The red and blue spectral response logarithmic ratio is: K rb =ln(K r / K b ); The red-green spectral response logarithmic ratio is: K rg =ln(K r / K g ); Among them, K r is the red channel spectral response coefficient, K g is the green channel spectral response coefficient, K b is the blue channel spectral response coefficient.

3. A temperature measurement method according to claim 2, characterized in that: The relationship between multiple brightness values ​​and temperatures constructed according to multiple spectral response logarithmic ratios and two-color temperature measurement formulas includes: The two-color temperature measurement formula is: Where C2 represents the second radiation constant; λ r , g , b Respectively represent the three primary color wavelengths of the temperature field image; T rg and T rb are the temperature measurement results of the red-green two-color temperature measurement method and the red-blue two-color temperature measurement method; R e , G e , B e Represent the brightness values ​​of the R channel, G channel, and B channel respectively; ε(λ,T) is a dimensionless parameter representing the spectral emissivity; The red-blue spectral response logarithmic ratio and the red-green spectral response logarithmic ratio are respectively characterized by the brightness values ​​of the three primary colors and a plurality of unknown parameters, and then the relationship between the plurality of brightness values ​​and the temperature is determined as follows: Among them, a rg’ 、b rg’ 、c rg’ 、a rb’ 、b rb’ and c rb’ is the unknown parameter.

4. A temperature measurement method according to claim 3, characterized in that: The neural network model for mapping the brightness value to the predicted temperature is constructed using the unknown parameters as network weights, including: Corresponding to the relationship between brightness value and temperature, a neural network model is constructed to map brightness value to predicted temperature, and the single-channel expression of the neural network model is obtained as follows: Among them, g1, g2 and g3 all represent activation functions, a rg 、b rg 、c rg 、a rb 、b rb and c rb represents the network weight to be solved corresponding to the unknown parameters, C rg and C rb represents a constant, T' rg and T' rb Respectively represent the output of a single channel; Corresponding to the number of the unknown parameters, a multi-channel neural network model is constructed according to the single-channel expression.

5. A temperature measurement method according to claim 1, characterized in that: The neural network model is trained by the sample temperature field image and physical constraints to determine the unknown parameters, and then the unknown parameters are substituted into the relationship between the brightness value and the temperature to obtain the temperature prediction model including: Inputting the sample temperature field image into the neural network model to obtain a predicted temperature; Determining a physical loss between a plurality of predicted temperatures according to the physical constraint, training the neural network model with the goal of minimizing the physical loss, and determining a value of the unknown parameter; The value of the unknown parameter is substituted into the relationship between the brightness value and the temperature, and a weight coefficient is determined according to a plurality of sample data to obtain a temperature prediction model.

6. A temperature measurement method according to claim 5, characterized in that: Determining the weight coefficient according to the plurality of sample data to obtain the temperature prediction model comprises: Substituting a plurality of sample data into a relational expression between brightness value and temperature for simultaneous calculation, and determining a weight coefficient of the relational expression between brightness value and temperature; Then, the weight coefficient is substituted into the relationship between the brightness value and the temperature to determine the temperature prediction model.

7. A temperature measuring device, characterized in that: The device comprises: An acquisition module is used to acquire a sample temperature field image and sample data corresponding to the real temperature; A construction module is used to extract the three-primary color brightness values ​​of the sample temperature field image, and to characterize a plurality of spectral response logarithmic ratios respectively through the three-primary color brightness values ​​and a plurality of unknown parameters, and then to construct a plurality of relationship equations between brightness values ​​and temperature according to the plurality of spectral response logarithmic ratios and a two-color temperature measurement method formula; the spectral response logarithmic ratio is the logarithmic ratio of the spectral response intensity of one primary color channel relative to another primary color channel; to construct equations between the relationship equations of the plurality of brightness values ​​and temperature, and to use the equations as physical constraints; to construct a neural network model that maps brightness values ​​to predicted temperatures using the unknown parameters as network weights; to train the neural network model through the sample temperature field image and the physical constraints, to determine the unknown parameters, and then to substitute the unknown parameters into the relationship equation between brightness values ​​and temperature to obtain a temperature prediction model; The acquisition module is also used to acquire a temperature field image of the target object; The prediction module is used to input the temperature field image of the target object into the temperature prediction model to obtain the temperature of the target object.

8. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

9. A computer device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method according to any one of claims 1 to 6 is implemented.

Images