An infrared thermal imaging temperature measurement method for a dynamic water mist interference scene

Abstract

An infrared thermal imaging temperature measurement method for dynamic water mist interference scene, which synchronously acquires infrared thermal images and visible light images of the measured object by using an infrared thermal imager and a visible light camera, and extracts the region of interest where the measured object is located based on the temperature difference between the region of interest and the background region, and then acquires the temperature value of the measured object. Considering the error caused by dynamic water mist to infrared temperature measurement, for the first time, starting from the color and texture features of the visible light image of the measured object, a channel difference contrast feature is proposed to quantify the dynamic water mist interference. For the first time, an intelligent compensation model is used to estimate the water mist transmittance in the visible light band and correct it to the infrared band, directly predict the temperature measurement error, and thus realize the compensation of the temperature measurement result. The problem of difficulty in mechanism modeling analysis and difficulty in directly and accurately detecting temperature of the infrared thermal imager under dynamic water mist interference is solved, and the temperature of the measured object under dynamic water mist interference is realized in real time and online high-precision detection.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of infrared thermal imaging temperature measurement, and particularly relates to an infrared thermal imaging temperature measurement method for a dynamic water mist interference scene. BACKGROUND

[0002] Temperature is a parameter that must be measured in many industrial processes or scientific experiments, and is important data support for temperature control, quality evaluation, and safety production. An infrared thermal imager has been widely applied to the steel, metallurgy, and power electronics industries due to its advantages of non-contact temperature measurement, no harmful radiation, and real-time online estimation. However, the emissivity of the surface of the measured object, the temperature measurement distance, the viewing angle, the transmittance, and other interference factors will all affect the temperature measurement accuracy of the infrared thermal imager.

[0003] At present, there are few studies on the influence of interference media between the infrared thermal imager and the measured object. However, water mist is a common interference medium between the infrared thermal imager and the measured object in industrial production processes. For example, the temperature measurement of engine blades, the surface temperature measurement of high-temperature steel billets, the surface temperature measurement of hot-rolled steel strips, and the rotary kiln in the sintering process are all affected by water mist interference factors, resulting in a large deviation in the temperature measurement results and limiting the application of the infrared thermal imager. Existing infrared temperature measurement methods for medium-type interference are mostly based on the radiation temperature measurement mechanism, and an infrared temperature measurement model for interference media is established. However, the parameters in the model need to be estimated, and it is difficult to directly apply the model to complex industrial temperature measurement scenes.

[0004] Therefore, the present application proposes an infrared thermal imaging temperature measurement method and system for a dynamic water mist interference scene. A temperature measurement system combining infrared vision and visible light vision is designed. Water mist-related features such as hue difference, multi-scale local maximum saturation, multi-scale local maximum brightness, and channel difference contrast are extracted from the visible light image of the measured object. The quantification of the dynamic water mist transmittance is realized. A correction model between the water mist transmittance in the visible light band and the water mist transmittance in the infrared band is constructed. An infrared temperature measurement intelligent compensation model is established to directly construct the mapping relationship between the water mist-related features and the infrared temperature measurement error. The compensation of the error in the infrared thermal imager temperature measurement result under the water mist interference is realized. The temperature measurement accuracy of the infrared thermal imager is improved, and the problem of limited application of the infrared thermal imager in the dynamic water mist interference scene is overcome.

[0005] The present application is different from the prior art as follows:

[0006] Compared with application number 202211583199.6:

[0007] CN 116046176A infrared thermal imaging temperature measurement error online compensation method and system for water mist interference

[0008] This invention proposes a method for compensating infrared thermal imaging temperature measurement errors caused by water mist interference. It estimates the water mist transmittance through dark channel a priori estimation, combines the Lambert-Beer law to determine the water mist transmittance in the infrared light band, and constructs a temperature compensation model to correct the temperature measurement error.

[0009] However, in fact, there are calculation errors in the water mist transmittance estimated by the dark channel prior. The Lambert-Beer law cannot accurately determine the water mist transmittance in the infrared light band, resulting in error accumulation, making it difficult to use for temperature compensation in dynamic water mist interference scenes.

[0010] Compared with application number 202310820871.7:

[0011] CN 117073847 A Temperature field online detection method and system based on multi-source vision under water mist interference

[0012] This invention utilizes a two-stage water mist transmittance field estimation method and establishes a detection model based on infrared temperature measurement principles to achieve temperature compensation.

[0013] This invention requires the prior measurement of the constant term in the relative water mist concentration to achieve online temperature compensation, and the compensation method based on the infrared temperature measurement principle can only be applied to scenarios where the water mist temperature is equal to the ambient temperature, making it difficult to achieve intelligent online compensation of infrared temperature measurement. Summary of the Invention

[0014] In order to solve the above technical problems, the present invention proposes an infrared thermal imaging temperature measurement method for dynamic water mist interference scenes. The temperature measurement method proposed in the present invention can compensate the temperature measurement results obtained by the infrared thermal imager when there is complex dynamic water mist interference in the optical path in real time online. There is no need to model the infrared radiation temperature measurement mechanism of water mist interference. It has the advantages of being intelligent, fast, and easy to implement.

[0015] To achieve the above object, the technical solution adopted by the present invention is:

[0016] An infrared thermal imaging temperature measurement method for dynamic water mist interference scenes includes the following steps:

[0017] (1) Using an infrared thermal imager and a visible light camera to synchronously acquire infrared thermal images and visible light images of the object under test, and performing multimodal image registration on the obtained infrared thermal images and visible light images;

[0018] (2) Based on the color and texture characteristics of visible light images, the water mist-related features such as hue difference, multi-scale local maximum saturation, multi-scale local maximum brightness and channel difference contrast are calculated to identify and quantify the dynamic water mist interference between the object under test and the infrared thermal imager optical path;

[0019] (3) Put forward two-stage infrared band water mist transmittance estimation method, first according to the dynamic water mist interference quantification feature, the water mist transmittance of visible light band is roughly estimated, then the forecast model with the input of the water mist transmittance of the measured object in the visible light band, saturation, brightness and channel difference contrast, and the output of the infrared band transmittance, the water mist transmittance in the visible light band is corrected to the infrared band;

[0020] (4) Build an infrared temperature measurement intelligent compensation model, which can automatically identify whether the infrared thermal imager temperature measurement result is disturbed by water mist, and compensate the temperature of the measurement result with error.

[0021] As a further improvement of the application, the specific process of step (1) is as follows:

[0022] a) Obtain the infrared thermal image and visible light image of the measured object, and calculate the temperature value of the region of interest;

[0023] Install the infrared thermal imager and visible light camera together in front of the measured object, maintain appropriate measurement distance and focal length so that the measured object is as much as possible in the middle of the infrared thermal image and visible light image and occupies a large field of view area;

[0024] Use the infrared thermal imager and visible light camera to synchronously obtain the infrared thermal image and visible light image of the measured object for subsequent infrared temperature measurement;

[0025] Based on the temperature difference between the region of interest and the background region, the following algorithm is proposed to locate the region of interest where the measured object is located and extract the temperature value;

[0026] Step1: Read the original infrared thermal image of the measured object, count the temperature histogram of the original infrared thermal image, and sort the temperature values in descending order, use Otsu method to calculate the segmentation threshold T of the infrared thermal image threshold , traverse the original infrared thermal image to retain the pixel points with temperature value greater than T threshold , consider that these pixel points form the region of interest of the measured object;

[0027] Step2: After locating the region of interest of the measured object, sort the temperature values in the region of interest in descending order, take the average temperature T avg in the upper half of the temperature distribution as the original temperature measurement value of the measured object;

[0028] b) Multi-modal image registration;

[0029] According to the infrared thermal image and visible light image pair obtained in a), a feature-based method is used for image registration, key point detection is used to identify points with significance and stability in the image, to provide a basis for subsequent feature extraction, feature description is used to encode the region around the detected key points, to generate representative feature vectors, so that they have comparability between the infrared thermal image and the visible light image, and after feature matching, the best matching feature point pair is found by comparing the feature descriptors in different images, and a homography matrix is calculated to obtain the transformed infrared thermal image and visible light image pair.

[0030] As a further improvement of the application, the specific process of step (2) is as follows:

[0031] a) Water mist interference identification;

[0032] A visible light camera is used to identify whether there is water mist interference in the light path, and the visible light image under the influence of water mist interference is I, and according to the hue difference and channel difference mean features, a reasonable threshold value is set to realize the identification of water mist interference;

[0033] 1) Hue difference;

[0034]

[0035] In the formula, c is the RGB channel of the visible light image; h is the hue channel; H(I) is the hue difference;

[0036] The hue difference represents whether there is water mist interference in a region of the image, which can be calculated by the difference between the hue channel of the original image and its semi-inverse image. When the pixels in the image are not affected by water mist interference, the hue difference changes little, and when the pixels in the image are affected by water mist interference, the hue difference changes greatly, which can be used to detect whether there is water mist interference;

[0037] 2) Channel difference mean;

[0038]

[0039] In the formula, max is the maximum value of each pixel position in the RGB channel of the visible light image I; min is the minimum value of each pixel position in the RGB channel of the visible light image I; mean is the mean value operation;

[0040] The channel difference mean is used to represent whether there is water mist interference in the current visible light image, which can be calculated by the mean value of the difference between the maximum channel and the minimum channel of the original image. When there is no water mist interference in the image, the channel difference mean is large, and when there is water mist interference in the image, the channel difference mean will be small and close to 0;

[0041] b) Water mist interference quantification:

[0042] When there is no water mist on the light path between the visible light camera and the measured object, the brightness and texture features of the measured object are not destroyed; when there is water mist on the light path between the visible light camera and the measured object, the visible light imaging is affected by the water mist, the reflected light of the measured object is affected by the suspended particles of the water mist, and the proportion of global atmospheric light participating in imaging increases, resulting in that the thicker the water mist area in the visible light image, the higher the brightness, the lower the saturation, and the more blurred the texture;

[0043] 1) Multi-scale channel difference;

[0044]

[0045] Ω (x) = max (min (R (x), G (x), B (x) ), min (R (x), G (x), B (x) ) ) - min (R (x), G (x), B (x) ), x∈Ω r (x) is a local block with x as the center and a radius of r; f r (x) is a multi-scale channel difference;

[0046] The multi-scale channel difference represents the high and low of the water mist concentration in a region of the image, and is a measurement parameter for describing the difference between the maximum channel and the minimum channel in the image. For a thin mist region, there is a significant difference between the maximum channel and the minimum channel, and the corresponding channel difference is large. For a thick mist region, due to the serious decline of contrast, the channel difference tends to 0, and different water mist concentration regions can be adaptively distinguished;

[0047] 2) Multi-scale dark channel;

[0048]

[0049] Ω (x) = min (R (x), G (x), B (x) ), x∈Ω r (x) is a local block with x as the center and a radius of r; D r (x) is a multi-scale dark channel;

[0050] The multi-scale dark channel represents the minimum value of all pixel colors in a local block of the image, and is a measurement parameter for describing the water mist content in a local region of the image. The size of the local block will affect the performance of the dark channel. If the local block is too large, it will cause excessive defogging, and if the local block is too small, it will not be able to play a feature description role;

[0051] 3) Multi-scale local gradient;

[0052]

[0053] Ω (x) = max (|R (x) - R (x + r) |, |G (x) - G (x + r) |, |B (x) - B (x + r) |), x∈Ω r (x) is a local block with x as the center and a radius of r; is a gradient image of the visible light image; T r (x) is a multi-scale local gradient;

[0054] Gradient represents the intensity change of a point in the image, which is a measurement parameter for describing the rate and direction of the pixel intensity change in the image, and can be obtained by derivation of the image and expressed in vector form. The greater the amplitude of the gradient, the greater the intensity change of a pixel in the image along the gradient direction. When the image is affected by water mist, the texture features in the visible light image change from complex and uneven distribution to relatively uniform distribution, and the gradient decreases;

[0055] 4) Multi-scale local maximum contrast;

[0056]

[0057] wherein, Ω r (x) is a local block with x as the center and a radius of r; Ω s (y) is a local block with y as the center and a radius of s; C r (x) is a multi-scale local maximum contrast;

[0058] Local contrast is the variance of pixel intensity in the local block of the image and the center pixel. The local maximum value is further calculated in the area to obtain the multi-scale local maximum contrast. When the image is affected by water mist, the visibility of the corresponding area decreases, and the local maximum contrast also decreases;

[0059] 5) Multi-scale local maximum saturation;

[0060]

[0061] wherein, Ω r (x) is a local block with x as the center and a radius of r; S r (x) is a multi-scale local maximum saturation;

[0062] Saturation represents the purity or intensity of color, which is a measurement parameter for describing the color brightness in HSV color space. The value range is 0 to 1. The greater the saturation, the greater the intensity of the color. When the image is affected by water mist, the color intensity of the visible light image will weaken due to the interference of the suspended particles of the water mist, and the saturation decreases;

[0063] 6) Multi-scale local maximum brightness;

[0064]

[0065] wherein, Ω r (x) is a local block with x as the center and a radius of r; L I is the brightness map of the visible light image; L r (x) is a multi-scale local maximum brightness;

[0066] Brightness represents the brightness of color, which affects the visual brightness of color, and its value range is 0 to 1, when the image is affected by water mist, the visible light image will change from normal brightness to very bright, and the local maximum brightness of the image will also increase;

[0067] 7) Channel difference contrast;

[0068] R c (x) = 255 - I c (x)(10)

[0069]

[0070] In the formula, R c (x) is the inverse image of the visible light image; Con(x) is the channel difference contrast;

[0071] The channel difference contrast is the ratio of the maximum channel difference of the image and the minimum channel difference to the maximum channel of the inverse image, and its value range is 0 to 1, when the image is affected by water mist, the brightness of the blurred area will increase, and the corresponding channel difference will also increase, at the same time, the maximum channel of the inverse image obtained by exchanging light and dark areas will become smaller, so the channel difference contrast will increase, and vice versa.

[0072] As a further improvement of the application, the specific process of step (3) is as follows:

[0073] a) Stack asymmetric autoencoder;

[0074] The mapping function used by the encoder can be represented by the following formula:

[0075]

[0076] In the formula, h is the encoded hidden layer; s1 and s2 are sigmoid activation functions; W1, W2, b1 and b2 are weight coefficients and intercept terms respectively; is the input variable;

[0077] The decoder reconstructs the original input by mapping function from the hidden layer h, and the mapping function used by the decoder can be represented by the following formula:

[0078]

[0079] In the formula, is the decoded reconstructed original input; s3 is a sigmoid activation function; W3 and b3 are weight coefficients and intercept terms respectively;

[0080] The asymmetric autoencoder optimizes the model parameters through the loss function L in the following formula:

[0081]

[0082] where m is the number of input variables.

[0083] The asymmetric autoencoder can be regarded as a three-layer neural network composed of an encoder and a decoder, and the size of the output layer is the same as that of the input layer. The asymmetric autoencoder is trained unsupervisedly by making the reconstructed output as close as possible to the input, thereby improving the feature extraction capability of the model.

[0084] A feature extraction layer is added when stacking the connection, so that the next asymmetric autoencoder can better learn the output of the previous asymmetric autoencoder, thereby improving the prediction capability of the model. The feature extraction layer can be represented by the following formula:

[0085] k=sigmoid(W4f in )(15)

[0086] f out =ReLU(k⊙(W5f in ))(16)

[0087] where k is a weight vector; is an element multiplication operation; W4 and W5 are weight coefficients; f in and f out are input feature vectors and output feature vectors, respectively;

[0088] For the hidden layer features output by the asymmetric autoencoder, two learnable and equal-size weight vectors are used to weight the input features, one branch uses a sigmoid function as the activation function, and the other branch is element-wise multiplied and then passed through a ReLU activation function to obtain the final output features.

[0089] b) Two-stage infrared band water mist transmittance estimation;

[0090] According to the dynamic water mist interference quantification method proposed in (2), due to the influence of water mist suspended particles in the air, the greater the water mist concentration, the higher the brightness and the lower the saturation in the corresponding area. Therefore, the fuzzy area in the fog image has the characteristics of high brightness and low saturation. Generally, the water mist concentration increases with the change of scene depth, and the above phenomenon can be represented by the following formula:

[0091] d(x)∝c(x)∝function(f r (x),D r (x),T r (x),C r (x),S r (x),L r (x),Con(x))(17)

[0092] where d(x) is the scene depth; c(x) is the water mist concentration;

[0093] And the transmittance and the scene depth have the following relationship:

[0094] t(x) = e -βd(x) (18)

[0095] where t(x) is the transmittance of the visible light band; d(x) is the scene depth; β is the atmospheric scattering coefficient;

[0096] Therefore, by quantifying the multi-scale channel difference, dark channel, gradient and other features, the transmittance of the visible light band can be roughly estimated;

[0097] In addition, according to the Lambert-Beer law, the mass extinction coefficient of water mist is related to the wavelength;

[0098] t wm (λ) = exp[-β'd'C'k'(λ)] (19)

[0099] where t wm (λ) is the transmittance of the visible light band; β' is the proportional coefficient; d' is the influence distance of the water mist; C' is the relative water mist concentration; k'(λ) is the mass extinction coefficient of the water mist;

[0100] As can be seen from equation (17), the saturation S and the brightness V of the visible light image have a relationship with most of the water mist concentration quantification features, and the channel difference contrast feature can quantify the water mist concentration. The transmittance of the water mist in the visible light band is corrected to the infrared band through the saturation, brightness and channel difference contrast features;

[0101] t inf (λ1) = function(t vis (λ2), S, V, Con) (20)

[0102] where t inf (λ1) is the transmittance of the infrared band; t vis (λ2) is the transmittance of the visible light band; λ1 and λ2 are the imaging bands of the infrared thermal imager and the visible light camera respectively; S is the saturation; V is the brightness; and Con is the channel difference contrast;

[0103] The saturation S and the brightness V are defined as follows:

[0104]

[0105] V = max(R, G, B) (22)

[0106] where R, G and B correspond to the red, green and blue color channels in the image respectively.

[0107] As a further improvement of the present application, the specific process of step (4) is as follows:

[0108] Step1: Collect the infrared thermal images and visible light images of the measured object under different temperatures, different scenes, and different water mist concentrations, simultaneously collect the true temperature of the measured object and the measured temperature affected by water mist, and calculate the water mist transmittance label true value in the infrared band according to the following formula:

[0109]

[0110] In the formula, t label is the water mist transmittance label true value; T effect , T truth and T u are the measured temperature affected by water mist, the true temperature of the measured object and the environment temperature respectively; n is a fitting coefficient related to the detector material of the infrared thermal imager;

[0111] Randomly extract two-thirds of the collected paired infrared thermal image and visible light image data set as the training set, and the remaining data set as the test set;

[0112] Step2: Establish a prediction model based on a stacked asymmetric autoencoder, extract the water mist concentration quantification features in the visible light image of the measured object, and convert the visible light image to the HSV color space, extract the saturation and brightness channels, and use the proposed two-stage infrared band water mist transmittance estimation method to obtain the infrared band water mist transmittance;

[0113] Step3: Input the infrared band water mist transmittance, saturation, brightness, channel difference contrast and measured temperature into the prediction model for training, and test the prediction effect of the model on the temperature measurement error on the test set;

[0114] Step4: Use the trained intelligent compensation model to compensate the measured temperature of the infrared thermal imager under the interference of water mist;

[0115] which is expressed by the following formula:

[0116]

[0117] ΔT = function (t inf , S, V, Con, T origin )(25)

[0118] In the formula, T and T origin are the output temperature and the original temperature measurement result respectively; ΔT is the prediction error;

[0119] The loss function constraint based on physical information ensures that the intelligent model can accurately estimate the water mist transmittance in the infrared band while also estimating the measurement error. The loss function of the intelligent compensation model composed of physical loss and label loss can be expressed as follows:

[0120]

[0121] L total =L label +L physi (28)

[0122] Where, L total , L label and L physi are the total loss function, label loss and physical loss respectively; m is the number of input variables; is the predicted water mist transmittance in the infrared band; T i ′、T i and T u are the affected measured temperature, true temperature and ambient temperature respectively; and are the predicted error and the true error respectively;

[0123] The collected infrared thermal images and visible light images are divided into data sets, and the established intelligent compensation model is trained using the training set. The temperature measurement error under water mist interference is predicted, thereby achieving accurate infrared temperature measurement under dynamic water mist interference.

[0124] Beneficial effects: Compared with the prior art, the technical solution of the present invention has the following beneficial technical effects:

[0125] (1) A channel difference contrast feature is proposed, which can calculate the channel difference and the ratio of the maximum channel of the inverted image based on the visible light image of the object to quantify the dynamic water mist interference in the optical path;

[0126] (2) A deep learning module based on stacked asymmetric autoencoders is proposed, and a feature extraction module based on one-dimensional convolution is combined with the stacked asymmetric autoencoder network to improve the prediction performance of the model;

[0127] (3) A two-stage infrared band water mist transmittance correction module based on features such as brightness, saturation, and channel difference contrast is proposed to correct the water mist transmittance in the visible light band and obtain the water mist transmittance in the infrared band online;

[0128] (4) Constructed the knowledge-guided infrared temperature measurement intelligent compensation model, a loss function constraint based on physical information is proposed, the physical loss and the label loss jointly constitute the loss function of the intelligent compensation model, while reducing the physical information loss, improve the prediction performance of the model on the infrared band water mist transmittance, thereby improve the prediction accuracy of the model on the temperature measurement error, realize the online accurate detection of temperature under dynamic water mist interference;

[0129] (5) An infrared thermal imaging temperature measurement method for dynamic water mist scene is proposed, which combines the advantages of infrared visual temperature measurement and visible light visual identification of water mist interference, and realizes accurate acquisition of the surface temperature of the measured object under dynamic water mist interference. BRIEF DESCRIPTION OF DRAWINGS

[0130] Figure 1 is the temperature measurement system of the present application;

[0131] Figure 2 is the temperature measurement method of the present application;

[0132] Figure 3 is the regression prediction result of the present application;

[0133] Figure 4 is the absolute error of the temperature measurement method of the present application before and after compensation;

[0134] Figure 5 is the temperature measurement result of the temperature measurement method of the present application before and after compensation. DETAILED DESCRIPTION

[0135] The technical solutions of the application will be further described in detail below in combination with the drawings. The described embodiments are only a part of the embodiments involved in the patent. All non-innovative embodiments of other researchers in the field on the embodiments belong to the protection scope of the patent.

[0136] The present application proposes an infrared thermal imaging temperature measurement method and system for dynamic water mist interference scene, Figure 1 is an infrared thermal imager and visible light camera intelligent cooperation temperature online detection system schematic diagram, which includes infrared thermal imager and visible light camera, etc. Figure 2 is the implementation step diagram of the infrared thermal imaging temperature measurement method proposed by the present application, which includes the following steps:

[0137] (1) The infrared thermal imager and visible light camera are used to synchronously acquire the infrared thermal image and visible light image of the measured object, and the obtained infrared thermal image and visible light image are subjected to multi-modal image registration.

[0138] (2) Based on the color and texture features of the visible light image, the hue difference, multi-scale local maximum saturation, multi-scale local maximum brightness, and channel difference contrast are calculated, which are related to the water mist, and the dynamic water mist interference between the measured object and the optical path of the infrared thermal imager is identified and quantified.

[0139] (3) A two-stage infrared band water mist transmittance estimation method is proposed. First, the water mist transmittance in the visible light band is roughly estimated according to the dynamic water mist interference quantification features, and then a prediction model is established with the water mist transmittance, saturation, brightness, and channel difference contrast in the visible light band as inputs and the infrared band transmittance as output, so as to correct the water mist transmittance in the visible light band to the infrared band.

[0140] (4) An infrared temperature measurement intelligent compensation model is constructed, which can automatically identify whether the temperature measurement result of the infrared thermal imager is disturbed by water mist, and compensate the temperature measurement result with error.

[0141] The specific implementation scheme is as follows:

[0142] (1) Obtain the infrared thermal image and the visible light image of the measured object, and locate the region of interest of the measured object for multi-modal image registration

[0143] a) Obtain the infrared thermal image and the visible light image of the measured object, and calculate the temperature value of the region of interest

[0144] The infrared thermal imager and the visible light camera are installed together in front of the measured object, and the appropriate measurement distance and focal length are maintained so that the measured object is as much as possible in the middle of the infrared thermal image and the visible light image and occupies a larger field of view area.

[0145] The infrared thermal imager and the visible light camera are used to synchronously obtain the infrared thermal image and the visible light image of the measured object for subsequent infrared temperature measurement.

[0146] Since the region of interest occupies a small proportion in the infrared thermal image, and because the resolution, installation position, and field of view angle of the visible light camera and the infrared thermal imager are all different, the rich color and texture information in the visible light image and the temperature value in the infrared thermal image are not one-to-one corresponding, so it is necessary to locate the region of interest of the measured object. Based on the temperature difference between the region of interest and the background region, the following algorithm is proposed to locate the region of interest of the measured object and extract the temperature value.

[0147] Step 1: Read the original infrared thermal image of the measured object, count the temperature histogram of the original infrared thermal image, and sort the temperature values in descending order. The threshold T of the infrared thermal image is calculated using the Otsu method. threshold , traverse the original infrared thermal image to retain the temperature value greater than T thresholdThe pixel points are considered to form a region of interest of the measured object.

[0148] Step 2: After locating the region of interest of the measured object, the temperature values in the region of interest are sorted in descending order, and the average temperature T in the upper half of the temperature distribution is taken as the temperature of the region of interest. avg as the original temperature measurement value of the measured object.

[0149] b) Multi-modal image registration

[0150] According to the infrared thermal image and visible light image pair obtained in a), a feature-based method is used for image registration. Key point detection is used to identify points with significance and stability in the image, providing a basis for subsequent feature extraction. Feature description is used to encode the region around the detected key points, generating representative feature vectors that are comparable between the infrared thermal image and the visible light image. After feature matching, the best matching feature point pairs are found by comparing the feature descriptors in different images. The homography matrix is calculated to obtain the transformed infrared thermal image and visible light image pair.

[0151] (2) Dynamic water mist interference recognition and quantification

[0152] a) Water mist interference recognition

[0153] Due to the lack of texture detail information in the infrared thermal image, it is difficult to identify whether the infrared thermal imager temperature measurement result is disturbed by water mist. Therefore, the present application proposes to use a visible light camera to identify whether there is water mist interference in the light path. Set the visible light image disturbed by water mist as I. According to the hue difference and channel difference mean features, a reasonable threshold can be set to realize the recognition of water mist interference.

[0154] 1) Hue difference

[0155]

[0156] In the formula, c is the RGB channel of the visible light image; h is the hue channel; and H(I) is the hue difference.

[0157] The hue difference represents whether there is water mist interference in a region of the image, which can be calculated by the hue channel difference between the original image and its semi-inverse image. When the pixels in the image are not disturbed by water mist, the hue difference changes little, and when the pixels in the image are disturbed by water mist, the hue difference changes greatly, which can be used to detect whether there is water mist interference.

[0158] 2) Channel difference mean

[0159]

[0160] Where max is the maximum value of each pixel position of the visible light image I on the RGB channel; min is the minimum value of each pixel position of the visible light image I on the RGB channel; mean is the mean operation.

[0161] The channel difference mean is used to indicate whether the current visible light image is affected by water fog. It is calculated by taking the mean of the difference between the maximum and minimum channels in the original image. When there is no water fog in the image, the channel difference mean is large. However, when there is water fog, the channel difference mean decreases to near zero.

[0162] b) Quantification of water mist interference

[0163] When there is no water mist on the optical path between the visible light camera and the object being measured, it is not affected by the water mist, and the brightness and texture characteristics of the object being measured are not destroyed. When there is water mist on the optical path between the visible light camera and the object being measured, the visible light imaging is affected by the water mist, and the reflected light of the object being measured is affected by the suspended particles in the water mist. The proportion of global atmospheric light participating in the imaging increases, resulting in higher brightness, lower saturation, and more blurred texture in areas with thicker water mist in the visible light image.

[0164] 1) Multi-scale channel differences

[0165]

[0166] Where, Ω r (x) is a local block with x as the center and radius r; f r (x) is the multi-scale channel difference.

[0167] Multi-scale channel difference characterizes the level of water mist concentration in a specific area of ​​an image. It is a metric used to describe the difference between the maximum and minimum channels in an image. In areas of light fog, there is a significant difference between the maximum and minimum channels, resulting in a large channel difference. In areas of dense fog, the channel difference approaches zero due to a significant drop in contrast, allowing for adaptive differentiation between areas of varying water mist concentrations.

[0168] 2) Multi-scale dark channel

[0169]

[0170] Where, Ω r (x) is a local block with x as the center and radius r; D r (x) is the multi-scale dark channel.

[0171] The minimum value of the multi-scale dark channel representing the color of all pixels in the local block of the image is a measurement parameter for describing the water mist content in the local area of the image. The size of the local block will affect the performance of the dark channel, and too large local block will cause excessive defogging, and too small local block cannot play a feature description role.

[0172] 3) Multi-scale local gradient

[0173]

[0174] In the formula, Ω r (x) is a local block with x as the center and a radius of r; is a gradient image of the visible light image; T r (x) is a multi-scale local gradient.

[0175] The gradient represents the intensity change of a point in the image, which is a measurement parameter for describing the intensity change rate and direction of the image, and can be obtained by derivation of the image and expressed in vector form. The greater the amplitude of the gradient, the greater the intensity change of a pixel in the image along the gradient direction. When the image is affected by water mist, the texture features in the visible light image change from complex and uneven distribution to relatively uniform distribution, and the gradient decreases.

[0176] 4) Multi-scale local maximum contrast

[0177]

[0178] In the formula, Ω r (x) is a local block with x as the center and a radius of r; Ω s (y) is a local block with y as the center and a radius of s; C r (x) is a multi-scale local maximum contrast.

[0179] The local contrast is the variance of the intensity of the pixels in the local block of the image and the center pixel, and the local maximum value is further calculated in the area to obtain the multi-scale local maximum contrast, which is a measurement parameter for describing the visibility of the image in the local area. When the image is affected by water mist, the visibility of the corresponding area decreases, and the local maximum contrast also decreases.

[0180] 5) Multi-scale local maximum saturation

[0181]

[0182] In the formula, Ω r (x) is a local block with x as the center and a radius of r; S r (x) is a multi-scale local maximum saturation.

[0183] Saturation represents the purity or intensity of color, which is a measurement parameter in HSV color space to describe the color vividness, and its value range is 0 to 1, the greater the saturation, the greater the intensity of the color. When the image is affected by water mist, due to the interference of water mist suspended particles, the color intensity of the visible light image will be weakened, and the saturation will be smaller.

[0184] 6) Multi-scale local maximum brightness

[0185]

[0186] In the formula, Ω r (x) is a local block with x as the center and a radius of r; L I is the brightness map of the visible light image; L r (x) is the multi-scale local maximum brightness.

[0187] Brightness represents the brightness of color, which affects the brightness of color in vision, and its value range is 0 to 1. When the image is affected by water mist, the visible light image will change from normal brightness to very bright, and the local maximum brightness of the image will also increase.

[0188] 7) Channel difference contrast

[0189] R c (x) = 255 - I c (x) (10)

[0190]

[0191] In the formula, R c (x) is the inverted image of the visible light image; Con(x) is the channel difference contrast.

[0192] The channel difference contrast is the ratio of the maximum channel difference and the minimum channel difference to the maximum channel of the inverse image, and its value range is 0 to 1. When the image is affected by water mist, due to the increase of the brightness of the blurred area, the corresponding channel difference will also increase, and at the same time, the maximum channel of the inverse image obtained by exchanging light and dark areas will become smaller, so the channel difference contrast will increase, and vice versa.

[0193] (3) Two-stage infrared band water mist transmittance estimation based on stacked asymmetric autoencoder

[0194] a) Stacked asymmetric autoencoder

[0195] The present application proposes a new autoencoder model structure, which is different from the previous symmetric encoder-decoder design method, wherein the mapping function used by the encoder can be represented by the following formula.

[0196]

[0197] where h is the encoded hidden layer; s1 and s2 are sigmoid activation functions; W1, W2, b1 and b2 are weight coefficients and intercept terms, respectively; is the input variable.

[0198] The decoder reconstructs the hidden layer h into the original input through a mapping function, which can be expressed as follows.

[0199]

[0200] where, is the decoded reconstructed original input; s3 is a sigmoid activation function; W3 and b3 are weight coefficients and intercept terms, respectively.

[0201] The asymmetric autoencoder optimizes the model parameters through a loss function L as follows.

[0202]

[0203] where m is the number of input variables.

[0204] The asymmetric autoencoder can be regarded as a three-layer neural network composed of an encoder and a decoder, and the size of the output layer is the same as that of the input layer. The asymmetric autoencoder performs unsupervised training on the encoder by making the reconstructed output as close as possible to the input, thereby improving the feature extraction capability of the model.

[0205] The stacked asymmetric autoencoder is a deep network stacked by multiple trained asymmetric autoencoders. It can learn more complex relationships between the input layer and the output layer, and obtain higher-order and more abstract features of the original input. After the first asymmetric autoencoder is trained, the output of the hidden layer is used as the input of the second asymmetric autoencoder, and the error backpropagation algorithm is used to train each layer in the above manner. A fully connected layer is added to the trained stacked asymmetric autoencoder, thereby establishing a deep neural network that can be used for data prediction.

[0206] In addition, unlike the simple way of directly stacking and connecting asymmetric autoencoders, the present application proposes adding a feature extraction layer when stacking and connecting, so that the next asymmetric autoencoder can better learn the output of the previous asymmetric autoencoder and improve the prediction capability of the model. The feature extraction layer can be expressed as follows.

[0207] k = sigmoid(W4f in )(15)

[0208] f out = ReLU(k ⊙ (W5f in )(16)

[0209] where k is a weight vector; is an element-wise multiplication operation; W4 and W5 are weight coefficients; f in and f out are input and output feature vectors, respectively.

[0210] For the hidden layer features of the asymmetric autoencoder output, two learnable and equal-size weight vectors are used to weight the input features, respectively, one branch uses a sigmoid function as the activation function, and the other branch is element-wise multiplied and then activated by the ReLU activation function to obtain the final output features.

[0211] b) Two-stage infrared band water mist transmittance estimation

[0212] According to the dynamic water mist interference quantification method proposed in (2), due to the influence of water mist suspended particles in the air, the greater the water mist concentration, the higher the brightness and the lower the saturation in the corresponding area, so the fuzzy area in the fog image has the characteristics of high brightness and low saturation. In addition, generally, the water mist concentration will increase with the change of scene depth, and the above phenomenon can be represented by the following formula:

[0213] d(x)∝c(x)∝function(f r (x),D r (x),T r (x),C r (x),S r (x),L r (x),Con(x))(17)

[0214] where d(x) is the scene depth; c(x) is the water mist concentration.

[0215] And the transmittance and scene depth have the following relationship:

[0216] t(x)=e -βd(x) (18)

[0217] where t(x) is the visible light band transmittance; d(x) is the scene depth; β is the atmospheric scattering coefficient.

[0218] Therefore, by quantifying the multi-scale channel difference, dark channel, gradient and other features, the visible light band water mist transmittance can be roughly estimated.

[0219] In addition, according to the Lambert-Beer law, the mass extinction coefficient of water mist is related to the wavelength.

[0220] t wm (λ)=exp[-β′d′C′k′(λ)](19)

[0221] In the formula, t wm (λ) is the water mist transmittance in the visible light band; β' is a proportional coefficient; d' is the influence distance of the water mist; C' is the relative water mist concentration; and k'(λ) is the mass extinction coefficient of the water mist.

[0222] The imaging bands of the visible light camera and the infrared thermal imager are not the same, and the visible light band water mist transmittance estimated from the visible light image cannot be directly used for temperature compensation and needs to be corrected to the infrared band. As can be seen from formula (17), the saturation S and the brightness V of the visible light image are related to most of the water mist concentration quantification features, and the channel difference contrast feature can quantize the water mist concentration, and the present application proposes to correct the water mist transmittance in the visible light band to the infrared band by the saturation, brightness and channel difference contrast features.

[0223] t inf (λ1) = function(t vis (λ2), S, V, Con) (20)

[0224] In the formula, t inf (λ1) is the water mist transmittance in the infrared band; t vis (λ2) is the water mist transmittance in the visible light band; λ1 and λ2 are the imaging bands of the infrared thermal imager and the visible light camera respectively; S is the saturation; V is the brightness; and Con is the channel difference contrast.

[0225] The saturation S and the brightness V are defined as follows:

[0226]

[0227] V = max(R, G, B) (22)

[0228] In the formula, R, G and B correspond to the red, green and blue color channels in the image respectively.

[0229] (4) Constructing an infrared temperature measurement intelligent compensation model

[0230] The visible light image and the infrared thermal image of the measured object disturbed by the water mist need to be manually photographed, and the dynamic water mist disturbance is simulated in the laboratory environment, so the amount of sample data available for learning is small. The stacked asymmetric autoencoder is an unsupervised learning method that can be used for small sample data and aims to extract complex features of high-dimensional data through layer-by-layer training. The infrared thermal imaging temperature measurement method for dynamic water mist disturbance scenes proposed in the present application can intelligently identify whether the current temperature measurement result is disturbed by the water mist according to the visible light image of the measured object, and intelligently compensate for the temperature measurement result with errors according to the color and texture features of the water mist disturbed area in the visible light image. The specific steps are as follows:

[0231] Step 1: Collect infrared thermal images and visible light images of the object under test at different temperatures, different scenes, and different water mist concentrations. At the same time, collect the actual temperature of the object under test and the measured temperature affected by water mist, and calculate the true value of the water mist transmittance label in the infrared band according to the following formula.

[0232]

[0233] Where, t label is the true value of the water mist transmittance label; T effect 、T truth and T u are the measured temperature affected by water mist, the real temperature of the measured object and the ambient temperature respectively; n is the fitting coefficient related to the detector material of the infrared thermal imager.

[0234] Two-thirds of the collected paired infrared thermal image and visible light image datasets are randomly selected as training sets, and the rest of the datasets are used as test sets.

[0235] Step 2: Establish a prediction model based on stacked asymmetric autoencoders to extract the quantitative features of water mist concentration in the visible light image of the object under test, convert the visible light image into the HSV color space, extract the saturation and brightness channels, and use the two-stage infrared band water mist transmittance estimation method proposed in (3) to obtain the infrared band water mist transmittance.

[0236] Step 3: Input the water mist transmittance, saturation, brightness, channel difference contrast and measured temperature in the infrared band into the prediction model for training, and test the model's prediction effect on temperature measurement error on the test set.

[0237] Step 4: Use the trained intelligent compensation model to compensate for the measured temperature of the infrared thermal imager under the interference of water mist.

[0238] The working mode of the infrared thermal imaging temperature measurement method and system for dynamic water mist interference scenes proposed in the present invention can be expressed by the following formula:

[0239]

[0240] ΔT=function(t inf ,S,V,Con,T origin )(25)

[0241] Where, T and T origin are the output temperature and the original temperature measurement results respectively; ΔT is the prediction error.

[0242] The application proposes a loss function constraint based on physical information, which ensures that the intelligent model accurately estimates the infrared band water mist transmittance while also estimating the measurement error, and the intelligent compensation model loss function composed of the physical loss and the label loss can be represented by the following formula:

[0243]

[0244] L total =L label +L physi (28)

[0245] In the formula, L total , L label and L physi are the total loss function, the label loss and the physical loss respectively; m is the number of input variables; is the predicted infrared band water mist transmittance; T i ', T i and T u are the affected measurement temperature, the real temperature and the environment temperature respectively; and are the prediction error and the real error respectively.

[0246] The collected infrared thermal images and visible light images are divided into data sets, the established intelligent compensation model is trained using the training set, and the temperature measurement error under the water mist interference is predicted, so as to realize accurate infrared temperature measurement under dynamic water mist interference.

[0247] Embodiment 1

[0248] In this embodiment, a high-temperature black body furnace is used to generate standard infrared radiation for model training, a metal cup filled with hot water is used as the experimental object, and the temperature measurement method of the application is applied to the object. The temperature measurement device is installed in front of the metal cup at a distance of 0.5m, and a temperature measurement system as shown in the accompanying Figure 1 is constructed.

[0249] The final prediction result is shown in Figure 3 , the absolute error graph is shown in Figure 4 , and the Figure 5 is the measured temperature of the metal cup before and after compensation using the temperature measurement method of the application. At the same time, a patch thermocouple is used to detect the actual temperature of the metal cup to verify the effectiveness of the temperature measurement method.

[0250] The above is only a preferred embodiment of the application, and does not limit the application in any other form, and any modification or equivalent change made according to the technical essence of the application still falls within the scope of the application claimed.

Claims

1. An infrared thermal imaging temperature measurement method for dynamic water mist interference scenes, characterized in that: The steps include: (1) Using an infrared thermal imager and a visible light camera to synchronously acquire infrared thermal images and visible light images of the object under test, and performing multimodal image registration on the obtained infrared thermal images and visible light images; (2) Based on the color and texture characteristics of visible light images, the features related to water mist, such as hue difference, multi-scale local maximum saturation, multi-scale local maximum brightness, and channel difference contrast, are calculated to identify and quantify the dynamic water mist interference between the object under test and the infrared thermal imager optical path; (3) A two-stage infrared band water mist transmittance estimation method is proposed. First, a rough estimate of the water mist transmittance in the visible light band is made based on the quantitative characteristics of dynamic water mist interference. Then, a prediction model is established with the water mist transmittance, saturation, brightness and channel difference contrast of the measured object in the visible light band as input and the infrared band transmittance as output. The water mist transmittance in the visible light band is corrected to the infrared band. (4) Construct an intelligent compensation model for infrared temperature measurement, which can automatically identify whether the temperature measurement results of the infrared thermal imager are interfered with by water mist, and perform temperature compensation for the erroneous measurement results; The specific process of step (4) is as follows: Step 1: Collect infrared thermal images and visible light images of the object under test at different temperatures, different scenes, and different water mist concentrations. At the same time, collect the actual temperature of the object under test and the measured temperature affected by water mist, and calculate the true value of the water mist transmittance label in the infrared band according to the following formula; （23） Where, is the true value of the water mist transmittance label; 、 and are the affected measured temperature, true temperature and ambient temperature respectively; is the fitting coefficient related to the detector material of the infrared thermal imager; Randomly extract two-thirds of the collected paired infrared thermal image and visible light image datasets as training sets, and the remaining datasets as test sets; Step 2: Establish a prediction model based on stacked asymmetric autoencoders to extract the quantitative features of water mist concentration in the visible light image of the object being measured and convert the visible light image into The saturation and brightness channels of the color space are extracted, and the water mist transmittance in the infrared band is obtained using the proposed two-stage infrared band water mist transmittance estimation method; Step 3: Input the infrared band's water mist transmittance, saturation, brightness, channel difference contrast, and measured temperature into the prediction model for training, and test the model's prediction effect on temperature measurement error on the test set; Step 4: Use the trained intelligent compensation model to compensate the temperature measured by the infrared thermal imager under the interference of water mist; It is expressed by the following formula: （24） （25） Where, and are the output temperature and the original temperature measurement results respectively; is the water mist transmittance; is the prediction error, is the saturation of the visible light image and is brightness, is the channel difference contrast; The loss function constraint based on physical information ensures that the intelligent model can accurately estimate the water mist transmittance in the infrared band while also estimating the measurement error. The loss function of the intelligent compensation model composed of physical loss and label loss can be expressed as follows: （26） （27） （28） Where, 、 and They are total loss function, label loss and physical loss respectively; is the number of input variables; is the predicted water mist transmittance in the infrared band; 、 and are the affected measured temperature, true temperature and ambient temperature respectively; and are the predicted error and the true error respectively; The collected infrared thermal images and visible light images are divided into data sets, and the established intelligent compensation model is trained using the training set. The temperature measurement error under water mist interference is predicted, thereby achieving accurate infrared temperature measurement under dynamic water mist interference.

2. The infrared thermal imaging temperature measurement method for dynamic water mist interference scenes according to claim 1 is characterized by: The specific process of step (1) is as follows: a) Obtain infrared thermal images and visible light images of the object being measured and calculate the temperature value of the area of ​​interest; Install the infrared thermal imager and visible light camera together in front of the object to be measured, maintain a suitable measuring distance and focal length so that the object to be measured is in the middle of the infrared thermal image and visible light image as much as possible and occupies a large field of view; Use an infrared thermal imager and a visible light camera to synchronously obtain infrared thermal images and visible light images of the object being measured for subsequent infrared temperature measurement; Based on the temperature difference between the region of interest and the background area, the following algorithm is proposed to locate the region of interest where the object is located and extract the temperature value; Step 1: Read the original infrared thermal image of the object to be measured, calculate the temperature histogram of the original infrared thermal image, sort it in descending order of temperature values, and use the Otsu method to calculate the segmentation threshold of the infrared thermal image , traverse the original infrared thermal image to retain the temperature value greater than The pixel points of the object are considered to be the area of ​​interest of the object under test. Step 2: After locating the area of ​​interest of the object being measured, sort the temperature values ​​in the area of ​​interest in descending order and take the average temperature in the upper half of the temperature distribution. As the original temperature measurement value of the object being measured; b) Multimodal image registration; Based on the infrared thermal image and visible light image pair obtained in a), a feature-based method is used for image registration. Key point detection is used to identify points in the image that are significant and stable, providing a basis for subsequent feature extraction. The areas around the detected key points are encoded in combination with feature descriptions to generate representative feature vectors. This ensures comparability and feature matching between the infrared thermal image and the visible light image. After that, the best matching feature point pair is found by comparing the feature descriptors in different images, and the homography matrix is ​​calculated to obtain the transformed infrared thermal image and visible light image pair.

3. The infrared thermal imaging temperature measurement method for dynamic water mist interference scenes according to claim 1 is characterized by: The specific process of step (2) is as follows: a) Water mist interference identification; Use a visible light camera to identify whether there is water mist interference in the light path. Assume that the visible light image interfered by water mist is ,According to the characteristics of hue difference and channel difference mean, ,setting a reasonable threshold can realize the recognition of water mist interference; 1) Hue difference; （1） （2） Where, is the RGB channel of the visible light image; is the hue channel; For hue difference; Hue difference indicates whether a certain area in the image is affected by water mist. It can be obtained by calculating the hue channel difference between the original image and its semi-inverse image. When the pixels in the image are not affected by water mist, the hue difference changes very little. However, when the pixels in the image are affected by water mist, the hue difference changes greatly. This can be used to detect whether there is water mist interference. 2) Channel difference mean; （3） In the formula, max is the visible light image The maximum value of each pixel position on the RGB channel; min is the visible light image The minimum value of each pixel position on the RGB channel; mean is the mean operation; The channel difference mean is used to indicate whether the current visible light image has water fog interference. It can be obtained by calculating the mean of the difference between the maximum channel and the minimum channel of the original image. When there is no water fog interference in the image, the channel difference mean is large, while when there is water fog interference in the image, the channel difference mean will become smaller and close to 0; b) Quantification of water mist interference: When there is no water mist on the optical path between the visible light camera and the object being measured, the light is not affected by the water mist, and the brightness and texture features of the object being measured are not destroyed. When there is water mist on the optical path between the visible light camera and the object being measured, the visible light imaging is affected by the water mist, and the reflected light of the object being measured is affected by the suspended particles in the water mist. The proportion of global atmospheric light participating in the imaging increases, resulting in higher brightness, lower saturation, and more blurred texture in areas with thicker water mist in the visible light image. 1) Multi-scale channel differences; （4） Where, For is the center and the radius is local blocks; is the multi-scale channel difference; Multi-scale channel difference characterizes the water mist concentration in a certain area of ​​the image. It is a metric used to describe the difference between the maximum and minimum channels in the image. For light fog areas, there is a significant difference between the maximum and minimum channels, and the corresponding channel difference is large. However, for dense fog areas, due to the severe drop in contrast, the channel difference tends to 0, which can adaptively distinguish areas with different water mist concentrations. 2) Multi-scale dark channels; （5） Where, For is the center and the radius is local blocks; It is a multi-scale dark channel; The multi-scale dark channel characterizes the minimum color value of all pixels in a local block of an image. It is a metric used to describe the water mist content in a local area of ​​the image. The size of the local block affects the characterization performance of the dark channel. If the local block is too large, it will lead to excessive dehazing, while if the local block is too small, it will not be able to serve as a feature description. 3) Multi-scale local gradients; （6） Where, is the gradient map of the visible light image; is the multi-scale local gradient; The gradient characterizes the intensity change of a certain point in the image. It is a metric used to describe the rate and direction of change of pixel intensity in the image. It can be obtained by taking the derivative of the image and expressed as a vector. The larger the gradient amplitude, the greater the intensity change of a certain pixel point in the image along the gradient direction. When the image is affected by water mist, the texture features in the visible light image change from a complex uneven distribution to a more uniform distribution, and the gradient decreases. 4) Multi-scale local maximum contrast; （7） Where, is the multi-scale local maximum contrast; Local contrast is the variance of the pixel intensity in a local block of the image and the central pixel. The local maximum is further calculated in this area to obtain the multi-scale local maximum contrast. When the image is affected by water fog, the visibility of the corresponding area decreases and the local maximum contrast also becomes smaller. 5) Multi-scale local maximum saturation; （8） Where, is the multi-scale local maximum saturation; Saturation represents the purity or intensity of the color. A metric parameter that describes the vividness of colors in color space. Its value range is 0 to 1. The greater the saturation, the greater the color intensity. When the image is affected by water mist, the color intensity of the visible light image will be weakened and the saturation will decrease due to the interference of suspended particles in the water mist. 6) Multi-scale local maximum brightness; （9） Where, is the brightness map of the visible light image; is the multi-scale local maximum brightness; Brightness represents the brightness of the color and affects the visual brightness of the color. Its value range is 0 to 1. When the image is affected by water mist, the visible light image will change from normal brightness to very bright, and the local maximum brightness of the image will also increase accordingly. 7) Channel difference contrast; （10） （11） Where, is the inverse image of the visible light image; is the channel difference contrast; Channel difference contrast is the ratio of the difference between the maximum and minimum channels of the image to the maximum channel of the inverse image. Its value range is 0 to 1. When the image is affected by water mist, the brightness of the blurred area increases, and the corresponding channel difference will also increase. At the same time, the maximum channel of the inverse image obtained by exchanging the bright and dark areas will become smaller, so the channel difference contrast will increase, and vice versa.

4. The infrared thermal imaging temperature measurement method for dynamic water mist interference scenes according to claim 1 is characterized by: The specific process of step (3) is as follows: a) Stacked asymmetric autoencoders; The mapping function used by the encoder is represented by the following equation; （12） Where, is the hidden layer after encoding; and for Activation function; 、 、 and are the weight coefficient and intercept term respectively; is the input variable; The decoder transforms the hidden layer into Reconstructing the original input, the mapping function used by the decoder can be expressed as follows; （13） Where, is the reconstructed original input after decoding; for Activation function; and are the weight coefficient and intercept term respectively; The asymmetric autoencoder uses the loss function in the following formula Optimize model parameters; （14） Where, is the number of input variables; The asymmetric autoencoder can be viewed as a three-layer neural network consisting of an encoder and a decoder. The size of the output layer is the same as the size of the input layer. The asymmetric autoencoder performs unsupervised training on the encoder by making the reconstructed output as equal to the input as possible, thereby improving the model's feature extraction capability. A feature extraction layer is added during the stacking connection so that the next asymmetric autoencoder can better learn the output of the previous asymmetric autoencoder and improve the prediction ability of the model. The feature extraction layer can be expressed as follows; （15） （16） Where, is the weight vector; is the element-wise multiplication operation; and is the weight coefficient; and are the input feature vector and the output feature vector respectively; For the hidden layer features output by the asymmetric autoencoder, two learnable and equal-sized weight vectors are used to weight the input features respectively. One branch uses the sigmoid function as the activation function, performs element-wise multiplication with the other branch, and then uses the ReLU activation function to obtain the final output features; b) Two-stage infrared band water mist transmittance estimation; According to the dynamic water mist interference quantification method proposed in (2), due to the influence of water mist suspended particles in the air, when the water mist concentration is greater, the brightness in the corresponding area is higher and the saturation is lower. Therefore, the blurred area in the foggy image has the characteristics of high brightness and low saturation. The water mist concentration will increase with the change of scene depth, which is expressed by the following formula: （17） Where, is the scene depth; is the water mist concentration; The relationship between transmittance and scene depth is as follows: （18） Where, is the transmittance in the visible light band; is the scene depth; is the atmospheric scattering coefficient; Therefore, a rough estimate of the water mist transmittance in the visible light band is made through multi-scale channel difference, multi-scale dark channel, multi-scale local gradient, multi-scale local maximum contrast, multi-scale local maximum saturation, multi-scale local maximum brightness and channel difference contrast; In addition, according to the Lambert-Beer law, the mass extinction coefficient of water mist is related to the wavelength; （19） Where, is the water mist transmittance in the visible light band; is the proportionality coefficient; is the impact distance of the water mist; is the relative water mist concentration; is the mass extinction coefficient of water mist; From formula (17), we can see that the saturation of the visible light image and brightness It is related to most of the quantitative features of water mist concentration, and the channel difference contrast feature can quantify the water mist concentration. The water mist transmittance in the visible light band is corrected to the infrared band through saturation, brightness and channel difference contrast features. （20） Where, is the water mist transmittance in the infrared band; is the water mist transmittance in the visible light band; and These are the imaging bands of the infrared thermal imager and the visible light camera respectively; is saturation; is brightness; Saturation and brightness The definition is as follows: （21） （22） Where, 、 and They correspond to the red, green, and blue color channels in the image respectively.

Images