A method for detecting leakage of flue gas from a refining furnace

By using area growth and grayscale value clustering combined with wind direction characteristics to update pixel grayscale values ​​in the refining furnace flue gas leakage detection, the problem of inaccurate contrast enhancement in the prior art is solved, and the accuracy and efficiency of detection are improved .

CN118608524BActive Publication Date: 2024-12-03WUXI DONGXONG HEAVY ARC-FURNACE CO LTD
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202411080976.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-08
Publication Date
2024-12-03
Estimated Expiration
2044-08-08

AI Technical Summary

Technical Problem

The prior art cannot accurately identify and distinguish the contrast between different dynamic highlight target areas and other areas, resulting in low accuracy of flue gas leakage detection for refining furnaces.

Method used

By acquiring continuous multi-frame infrared grayscale images, using area growth and grayscale value clustering to determine the temperature variable area block, compute the smoke scores based on wind direction characteristics, update the pixel grayscale values ​​to enhance contrast, and input the pre-trained neural network model for Test.

Benefits of technology

It improves the accuracy and processing efficiency of flue gas leakage detection in refining furnaces, can accurately distinguish the contrast between the flue gas area and other areas, and enhances the accuracy of the detection.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118608524B_ABST
    Figure CN118608524B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of image data processing technology, and more specifically, to a method for detecting a flue gas leakage in a refining furnace, the method comprising: obtaining a reference image in an infrared grayscale image collected around the refining furnace, determining a growth region with each pixel point in the reference image as a center point and a category of the growth region, determining a variable temperature region block of each category according to a number of pixels in the growth region of each category, and then calculating the score of the matching region block of each category of the variable temperature region block in each reference image belonging to the flue gas, updating the grayscale value of the pixel points in the matching region block of the corresponding category of the variable temperature region block, and obtaining a target infrared grayscale image, so as to input the target infrared grayscale image into a pre-trained neural network model to detect the leakage of the flue gas. The present invention can accurately enhance the contrast between the region where the flue gas is located and other regions in the infrared image, thereby improving the accuracy of the flue gas leakage detection.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of image data processing, and more particularly to a method for detecting a refining furnace fume leakage. Background Art

[0002] Refinery flue gas leak detection is an important part of industrial safety and environmental protection, especially in the steel, petrochemical and other industries, where flue gas leaks may lead to resource waste, environmental pollution and even safety accidents. At present, with the development of industrial automation and intelligence, refinery flue gas leak detection technology is also constantly improving to improve the accuracy and efficiency of detection.

[0003] In the related art, a patent application document with publication number CN115908195A discloses an adaptive contrast enhancement method for infrared images. The method includes: firstly, extracting the information of the original infrared image in real time for histogram statistics; secondly, obtaining the actual mapping grayscale range of the original infrared image based on the histogram statistics, intercepting and filtering out the bad pixels at both ends; thirdly, performing linear mapping dimming and platform histogram equalization dimming respectively; finally, performing weighted mixed dimming based on the two dimming results, obtaining an infrared image after adaptive contrast enhancement, automatically changing the contrast of the target and the background, and improving the environmental adaptability of the dynamic infrared image.

[0004] However, the scheme recorded in the application document only considers the enhancement of the contrast between the dynamic highlight target and the background, and is unable to distinguish the types of dynamic highlight targets. That is, when there are multiple dynamic highlight targets in the infrared image, it is impossible to enhance the contrast between the areas where different dynamic highlight targets are located and other areas in a differentiated manner, so that the infrared grayscale image enhanced based on the above scheme may not be able to accurately identify the target, which may reduce the accuracy of target recognition to a certain extent. Summary of the invention

[0005] In order to solve the problem that the contrast between the area where different dynamic highlight targets are located and other areas cannot be enhanced differently, resulting in low accuracy of target recognition, the present invention proposes a refinery flue gas leakage detection method.

[0006] According to one aspect of an embodiment of the present invention, a method for detecting a refining furnace flue gas leakage is provided, the method comprising the following steps:

[0007] Acquire a plurality of reference images from a plurality of consecutive frames of infrared grayscale images acquired around the refining furnace;

[0008] Taking each pixel in each reference image as the center point, regional growth is performed to obtain the growth region of each pixel, and the growth regions with overlapping parts are regarded as a type of growth region. In each type of growth region, a number of pixels whose ratio between the number of the growth regions and the total number of the growth regions in each type is greater than a preset value are screened to obtain the temperature change region block of each type in each reference image;

[0009] Determine the matching area blocks of any category of variable temperature area blocks in each reference image, and determine the score of the matching area block belonging to smoke according to the mean value of the deviation between the diffusion direction of the matching area block and the wind direction at the acquisition time of the corresponding reference image. The score is negatively correlated with the mean value, and the diffusion direction is the direction of the line connecting the two pixels with the farthest distance in the corresponding matching area block;

[0010] The grayscale value of the pixel in the matching area block is updated. The updated grayscale value is positively correlated with the score, and the target infrared grayscale image is obtained. The target infrared grayscale image is input into the pre-trained neural network model to detect smoke leakage.

[0011] The present invention can update the grayscale values ​​of the pixels in the matching area blocks of the variable temperature area blocks of the corresponding categories according to the scores of the matching area blocks of the variable temperature area blocks of each category in each reference image belonging to the flue gas, and can enhance the contrast between the variable temperature area blocks of each category and other areas in a differential manner. The higher the score belonging to the flue gas, the higher the contrast between the corresponding variable temperature area block and other areas. Therefore, the contrast between the area where the flue gas is located and other areas can be accurately enhanced, thereby improving the accuracy of the detection of flue gas leakage in the refining furnace.

[0012] In an exemplary embodiment of the present invention, a method for obtaining a reference image includes:

[0013] The grayscale value of the pixel point at the same pixel position in each infrared grayscale image is used as the ordinate, and the frame number of each infrared grayscale image is used as the abscissa to construct a grayscale change scatter plot;

[0014] Obtain the mean of the absolute values ​​of the slopes of the lines connecting adjacent points in the grayscale change scatter plot, the variance of the ordinates of each point, and the difference between the maximum ordinate value and the minimum ordinate value, and use the product of the mean, the variance and the difference as the degree of temperature change at the corresponding pixel position;

[0015] The pixel points at the pixel positions whose corresponding temperature change degree is greater than or equal to the preset threshold are taken as reference pixels, and the grayscale values ​​of each reference pixel are clustered to obtain two grayscale categories of the reference pixel points. The infrared grayscale image corresponding to the category with the larger grayscale mean is taken as the reference image.

[0016] The present invention can accurately screen out infrared grayscale images with large grayscale value changes, thereby performing analysis based on infrared grayscale images with large grayscale value changes, reducing the amount of processing for each frame of acquired infrared grayscale images, thereby improving processing efficiency.

[0017] In an exemplary embodiment of the present invention, the score of the matching area block belonging to smoke satisfies the following relationship:

[0018] ;

[0019] in, The score of the matching area block of any category of temperature-varying area block belonging to flue gas; The number of matching area blocks of the variable temperature area block of this category; This is the first temperature change area block of this category. The diffusion direction of the matching area blocks; The wind direction at the time of acquisition of the reference image corresponding to the matching area block; for the summation symbol; is the absolute value symbol; The natural constant An exponential function with base .

[0020] The present invention utilizes the feature that the diffusion direction of the smoke area is affected by the wind direction, and can accurately calculate the score of the matching area blocks of each category of temperature-changing area blocks belonging to the smoke.

[0021] In an exemplary embodiment of the present invention, a method for acquiring an infrared grayscale image of a target includes:

[0022] Calculate the gamma coefficient of the matching area block , To match the score of the area block belonging to smoke, The maximum value of the scores of the matching area blocks corresponding to all categories of temperature-changing area blocks belonging to flue gas;

[0023] The gamma coefficient is used to update the grayscale value of the pixel in the matching area block to obtain the target infrared grayscale image. The updated grayscale value is negatively correlated with the gamma coefficient.

[0024] The present invention can update the grayscale values ​​of the pixels in the matching area blocks of the variable temperature area blocks of the corresponding categories according to the scores of the matching area blocks of the variable temperature area blocks in each reference image belonging to the smoke, thereby accurately enhancing the contrast between the area where the smoke is located and other areas.

[0025] In an exemplary embodiment of the present invention, the grayscale value of the pixel in the matching area block is updated using the gamma coefficient to satisfy the following relationship:

[0026] ;

[0027] In the formula, For the The first matching area block The updated gray value of each pixel; For the The first matching area block The gray value of each pixel before updating; For the The pixel value of the reference image corresponding to the matching area block; For the Gamma coefficient of the matching area block.

[0028] In an exemplary embodiment of the present invention, a method for obtaining a matching region block includes:

[0029] Arrange the reference images in order of frame numbers, take any variable temperature region block in the current frame reference image as the target region block, obtain the variable temperature region block with the greatest matching degree with the target region block in the next frame reference image, and obtain the matching region block in the next frame reference image, wherein the matching degree reflects the difference between the area and position of the variable temperature region block in the next frame image and the target region block;

[0030] The matching region block in the next frame of the reference image is used as a new target region block for matching, and so on, to obtain the matching region block of the target region block in each frame of the reference image.

[0031] In an exemplary embodiment of the present invention, the matching degree satisfies the following relationship:

[0032] ;

[0033] In the formula, For the In the frame reference image, The matching degree between the temperature-varying area block and the target area block; For natural numbers An exponential function with base ; , are the area of ​​the target region and The area of ​​the temperature-changing region block; , are the horizontal coordinates of the central pixel of the target area block and the The horizontal coordinate of the central pixel point of the temperature-varying area block; , are the ordinates of the center pixel of the target area block, and The vertical coordinate of the central pixel of the temperature-varying area block; is the absolute value symbol.

[0034] In an exemplary embodiment of the present invention, a method for acquiring a growth area of ​​each pixel point includes:

[0035] Taking any pixel point in the reference image as the target pixel point, determining all target neighborhood pixel points within the neighborhood of the target pixel point whose grayscale value difference with the target pixel point is less than a first preset value, to form an initial growth area of ​​the target pixel point;

[0036] Acquire all pixel points within the neighborhood of each target neighborhood pixel point that meet the set conditions and are within the preset growth range, obtain all growth pixel points in the initial growth area, and constitute the growth area of ​​the target pixel point; wherein the set conditions are: the difference in grayscale value with the target pixel point is less than a first preset value, and the difference in grayscale value with the corresponding target neighborhood pixel point is less than a second preset value, and the first preset value is greater than the second preset value.

[0037] The present invention can accurately screen out pixel points with similar grayscale values ​​by setting two threshold conditions, thereby accurately determining the growth area of ​​each pixel point.

[0038] The present invention has the following beneficial effects:

[0039] The present invention determines the temperature variable area block of each category by integrating various data, so as to ensure the accuracy of the determined temperature variable area block, and utilizes the feature that the diffusion direction of the temperature variable area where the flue gas is located is affected by the wind direction to determine the score of the matching area block of each category of the temperature variable area block belonging to the flue gas, so as to ensure the accuracy of the determined score of the flue gas, and then utilizes the score of the matching area block of each category of the temperature variable area belonging to the flue gas to update the grayscale value of the pixel points in the matching area block of the corresponding category of the temperature variable area block, so as to enhance the contrast between the temperature variable area block of each category and other areas differently, and the higher the score belonging to the flue gas, the larger the corresponding updated grayscale value, so that the contrast between the temperature variable area block with a higher score belonging to the flue gas and other areas is relatively larger, so that the contrast between the area where the flue gas is located and other areas can be accurately enhanced, thereby improving the accuracy of the detection of flue gas leaks in the refining furnace. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood by reading the detailed description below with reference to the accompanying drawings. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0041] Figure 1 The present invention is a flowchart of a method for detecting a refining furnace flue gas leakage according to an embodiment of the present invention. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0043] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0044] See also Figure 1 , a refinery flue gas leakage detection method, comprising steps S1 to S4, specifically as follows:

[0045] S1: Acquire multiple reference images from a plurality of consecutive frames of infrared grayscale images acquired around the refining furnace.

[0046] In an exemplary embodiment of the present invention, the infrared grayscale image refers to an infrared image collected around the refining furnace by an infrared camera, which is used as an image basis to analyze the change of the grayscale value of each pixel. Specifically, 300 frames of infrared grayscale images can be collected continuously, and then the change of the grayscale value of each pixel in the continuous 300 frames of infrared grayscale images is analyzed, so as to make further judgments based on the analysis results. It should be noted that the infrared image collected by the infrared camera is a grayscale image, and this embodiment does not specifically limit the number of frames of the collected infrared grayscale image.

[0047] The reference image refers to an infrared grayscale image with pixels having large grayscale value changes, which is used as the image basis for further analysis. The reference image can be an infrared grayscale image of all acquired frames, or an image obtained by screening according to the grayscale value changes of each pixel in multiple consecutive frames of infrared grayscale images. This embodiment does not specifically limit the determination of the reference image.

[0048] In an exemplary embodiment of the present invention, the determination of the reference image may be achieved by the following steps:

[0049] The grayscale value of the pixel point at the same pixel position in each infrared grayscale image is used as the ordinate, and the frame number of each infrared grayscale image is used as the abscissa to construct a grayscale change scatter plot; the mean of the absolute value of the slope of the connecting line of adjacent points in the grayscale change scatter plot, the variance of the ordinate of each point, and the difference between the maximum value of the ordinate and the minimum value of the ordinate are obtained, and the product of the mean, the variance and the difference is used as the temperature change degree of the corresponding pixel point; the pixel points whose corresponding temperature change degree is greater than or equal to the preset threshold are used as reference pixel points, the grayscale values ​​of each reference pixel point are clustered, and two grayscale categories of the reference pixel points are obtained, and the infrared grayscale image corresponding to the category with the larger grayscale mean is used as the reference image.

[0050] It should be noted that when collecting multiple consecutive frames of infrared grayscale images, the infrared camera needs to be fixed so that the grayscale values ​​of the background part of the collected infrared grayscale images are basically similar in the multiple consecutive frames. When the temperature of the refining furnace changes, or the operator enters the shooting range, or there is a flue gas leakage in the refining furnace, the grayscale values ​​of some pixels in the infrared grayscale image will change significantly. Therefore, when detecting flue gas leakage in the refining furnace, only the infrared grayscale images with larger grayscale changes can be analyzed, thereby reducing the amount of calculation and improving detection efficiency.

[0051] Among them, the grayscale change scatter plot refers to an image that can reflect the grayscale value changes of all pixels at each pixel position in the infrared grayscale image, and is used to determine the temperature changes corresponding to each pixel; the temperature change degree refers to data that measures the severity of the grayscale changes of each pixel, and is used to determine the pixels with larger grayscale value changes, so as to determine all reference pixels.

[0052] Specifically, the temperature change degree of each pixel satisfies the following relationship:

[0053] ;

[0054] in, is the degree of temperature change at any pixel position; , are the maximum and minimum values ​​of the ordinates of each point in the grayscale change scatter diagram of the pixel position; is the total number of frames of the infrared grayscale image. The value of can be 300; is the grayscale change scatter plot of the pixel position. point as the left endpoint, The slope of the line segment determined by the points as the right endpoint; is the variance of the ordinate of each point in the grayscale change scatter plot of the pixel position.

[0055] in, It reflects the variation range of the grayscale values ​​of all pixels at this pixel position in 300 consecutive image frames. The smaller the value is, the smaller the grayscale value of all pixels at this pixel position in 300 consecutive image frames changes or remains basically unchanged, indicating that the temperature variation degree of this pixel position is small.

[0056] Optionally, the preset threshold can be set to 50. When the temperature change degree of any pixel position is greater than or equal to 50, all pixel points at this pixel position can be used as reference pixel points. When the temperature change degree of any pixel position is less than 50, it can be considered that the grayscale value of the pixel point at this pixel position changes little or remains basically unchanged, so that all pixel positions with large grayscale changes can be screened out, and then all reference pixel points can be determined. This embodiment does not specifically limit the size of the preset threshold.

[0057] Furthermore, the grayscale values ​​of the reference pixels in multiple consecutive frames of infrared grayscale images can be clustered to obtain two grayscale categories for all reference pixels, and then the infrared grayscale image corresponding to the category with the larger grayscale mean value can be used as the reference image. It should be noted that clustering the grayscale values ​​of each pixel is an existing technology and will not be described in detail here.

[0058] S2: Perform regional growth with each pixel in each reference image as the center point to obtain the growth area of ​​each pixel, take the growth areas with overlapping parts as a type of growth area, and select several pixels in each type of growth area whose ratio between the number of growth areas and the total number of growth areas in each type is greater than a preset value to obtain the temperature variable area block of each category in each reference image.

[0059] In an exemplary embodiment of the present invention, a growth region refers to a region obtained by performing region growth using a region growing algorithm with a certain pixel point as the center, and is used to determine the region where pixels with similar grayscale values ​​are located. By treating the growth regions with overlapping parts as a type of growth region, pixels with similar grayscale values ​​can be placed in the same region. It should be noted that the process of performing region growing using a region growing algorithm is a prior art and will not be described in detail herein.

[0060] The variable temperature area block refers to the area block composed of all pixels with similar grayscale values, which is used to locate the area where the temperature changes in the reference image. By screening a number of pixels in each type of growth area, where the ratio between the number of growth areas in which they are located and the total number of growth areas in each type is greater than a preset value, the influence of noise and other factors can be avoided, thereby ensuring the accuracy of the area composed of pixels with similar grayscale values, that is, ensuring the accuracy of the variable temperature area block of each category.

[0061] Optionally, the preset value can be set to 0.3. If, in any type of growth area, the ratio between the data of the growth area where any pixel point is located and the total number of growth areas of this type is greater than 0.3, the grayscale value of any pixel point can be close to the overall grayscale level of this type of growth area, so that the pixel point can be regarded as a pixel point in this type of growth area, so that all pixels belonging to this type of growth area can be screened to obtain the variable temperature area block of this category, and then determine the variable temperature area block of each category in each reference image.

[0062] In an exemplary embodiment of the present invention, the growth area of ​​each pixel can be determined by the following steps:

[0063] Any pixel point in the reference image is taken as the target pixel point, and all target neighborhood pixel points whose grayscale value difference with the target pixel point is less than a first preset value within the neighborhood range of the target pixel point are determined to constitute the initial growth area of ​​the target pixel point; all pixel points within the neighborhood range of each target neighborhood pixel point that meet the set conditions and are within the preset growth range are obtained to obtain all growth pixel points in the initial growth area to constitute the growth area of ​​the target pixel point; wherein the set conditions are: the grayscale value difference with the target pixel point is less than the first preset value, and the grayscale value difference with the corresponding target neighborhood pixel point is less than the second preset value, and the first preset value is greater than the second preset value.

[0064] Specifically, the first preset value can be set to 40, the second preset value can be set to 30, and the preset growth range can be set to 20×20; The pixel point is the target pixel point, and the The process of growing a region with a pixel as the center point includes: Within the eight neighborhoods of pixels, The gray value of the pixel is If the difference between the grayscale values ​​of pixels is less than 40, pixel points as the target neighborhood pixel points, thereby determining the All target neighboring pixels within the eight neighborhoods of the pixel are based on the pixel points and all corresponding target neighborhood pixels, determine the The initial growth area of ​​pixels.

[0065] Furthermore, when The pixel point in this initial growth area is When the pixel is within the eight-neighborhood range, if the Pixels and The difference between the grayscale values ​​of the pixels is less than 40, The difference between the grayscale values ​​of pixels is less than 30, and the The pixels are in the The pixel point is the center point of the 20×20 area, then the pixel points as the growth pixel points of the initial growth area, thereby determining all the growth pixel points of the initial growth area, and obtaining the first pixel point consisting of the initial growth area and all the corresponding growth pixel points The growth area of ​​​​pixels.

[0066] S3: Determine the matching area blocks of any category of variable temperature area blocks in each reference image, and determine the score of the matching area block belonging to smoke according to the mean of the deviation between the diffusion direction of the matching area block and the wind direction at the acquisition time of the corresponding reference image. The score is negatively correlated with the mean, and the diffusion direction is the direction of the line connecting the two pixels with the farthest distance in the corresponding matching area block.

[0067] It should be noted that when smoke leaks from the refining furnace, the diffusion direction of the smoke may be affected by the wind direction, while the temperature changes of the refining furnace itself and the movement of the operator will not be affected by the wind direction. Therefore, this feature can be used to determine the possibility that the temperature-varying area block in the reference image belongs to smoke, that is, the score belonging to smoke.

[0068] Optionally, an anemometer may be used to measure the wind direction of the workshop, so that the wind direction at the time of acquisition of each reference image can be determined.

[0069] In an exemplary embodiment of the present invention, the following steps can be used to determine the matching area blocks of any category of temperature-varying area blocks in each reference image:

[0070] Arrange the reference images in order of frame numbers, take any variable temperature region block in the current frame reference image as the target region block, obtain the variable temperature region block with the greatest matching degree with the target region block in the next frame reference image, and obtain the matching region block in the next frame reference image. The matching degree reflects the difference between the area and position of the variable temperature region block in the next frame image and the target region block; match the matching region block in the next frame reference image as a new target region block, and so on, to obtain the matching region block of the target region block in each frame reference image.

[0071] Optionally, the area of ​​the variable temperature region block may be represented by the number of all pixel points in the variable temperature region block, and the position of the variable temperature region block may be represented by the coordinates of the central pixel point in the variable temperature region block.

[0072] Specifically, the matching degree between the next frame reference image and the target area block satisfies the following relationship:

[0073] ;

[0074] In the formula, For the In the frame reference image, The matching degree between the temperature-varying area block and the target area block; For natural numbers An exponential function with base ; , are the area of ​​the target block and The area of ​​the temperature-changing region block; , are the horizontal coordinates of the central pixel of the target area block and the The horizontal coordinate of the central pixel point of the temperature-varying area block; , are the ordinates of the center pixel of the target area block, and The vertical coordinate of the central pixel of the temperature-varying area block; is the absolute value symbol.

[0075] It should be noted that when The smaller the In the frame reference image, The area of ​​the variable temperature area block is closer to the area of ​​the target area block; The smaller the In the frame reference image, The closer the central pixel point of the temperature-varying area block is to the central pixel point of the target area block, and The smaller, The smaller the corresponding The larger the The frame reference image The more temperature-varying area blocks there are, the better the match with the target area block.

[0076] Among them, when determining the After finding the temperature-varying region block with the greatest matching degree with the target region block in the frame reference image, the temperature-varying region block can be used as the new target region block, and then the first The matching degree between each temperature-varying region block in the frame reference image and the new target region block is calculated, so that the target region block is The matching area blocks in the frame reference image are obtained by analogy, and the matching area blocks of the target area blocks in each reference image are obtained, and then the matching area blocks of the variable temperature area blocks of each category in each reference image are determined.

[0077] Further, after determining the matching area blocks of each category of the variable temperature area blocks in each reference image, the score of the matching area blocks of any category of the variable temperature area blocks belonging to the smoke can be calculated. Specifically, the score of the matching area blocks of any category of the variable temperature area blocks belonging to the smoke satisfies the following relationship:

[0078] ;

[0079] In the formula, The score of the matching area block of any category of temperature-varying area block belonging to flue gas; The number of matching area blocks of the variable temperature area block of this category; This is the first temperature change area block of this category. The diffusion direction of the matching area blocks; The wind direction at the time of acquisition of the reference image corresponding to the matching area block; for the summation symbol; is the absolute value symbol; The natural constant An exponential function with base .

[0080] In another embodiment, the score of any category of temperature-changing area blocks belonging to flue gas may be calculated by other formulas, such as by directly taking negative values ​​instead of using an exponential function calculation formula.

[0081] S4: Update the grayscale value of the pixel points in the matching area block. The updated grayscale value is positively correlated with the score, and the target infrared grayscale image is obtained. The target infrared grayscale image is input into the pre-trained neural network model to detect smoke leakage.

[0082] In an exemplary embodiment of the present invention, the target infrared grayscale image is a reference image with a large contrast between the area where the flue gas is located and other areas, and is used as a basic image for detecting flue gas leakage in a refining furnace. The number of frames of the target infrared grayscale image is the same as the number of frames of the reference image, and this embodiment does not specifically limit the number of frames of the target infrared grayscale image.

[0083] In an exemplary embodiment of the present invention, the determination of the target infrared grayscale image can be achieved by the following steps:

[0084] Calculate the gamma coefficient of the matching area block , To match the score of the area block belonging to smoke, The maximum value of the score of the matching area block corresponding to all categories of variable temperature area blocks belonging to smoke; the gamma coefficient is used to update the grayscale value of the pixel points in the matching area block to obtain the target infrared grayscale image, and the updated grayscale value is negatively correlated with the gamma coefficient.

[0085] Specifically, the gamma coefficient is used to update the grayscale value of the pixel in the matching area block to obtain the target infrared grayscale image, which satisfies the following relationship:

[0086] ;

[0087] In the formula, For the The first matching area block The updated gray value of each pixel; For the The first matching area block The gray value of each pixel before updating; For the The pixel value of the reference image corresponding to the matching area block, in this embodiment, ; For the The gamma coefficient of the matching area block, .

[0088] It should be noted that The larger the corresponding The smaller, for The greater the enhancement amplitude, the corresponding gamma transformation The larger the value, the higher the score of the matching area block of any category of variable temperature area block belonging to flue gas, the greater the enhancement amplitude of the grayscale value of the pixel points within the matching area block of the variable temperature area block of that category, thereby enhancing the contrast between the flue gas and other areas, and further accurately detecting the flue gas leakage of the refining furnace.

[0089] In another embodiment, the matching area blocks of any category of variable temperature area blocks may be utilized in other ways to update the grayscale values ​​of pixels within the matching area blocks of the corresponding category of variable temperature area blocks, such as by a product calculation formula instead of a gamma change calculation formula.

[0090] Furthermore, before the target infrared grayscale image is input into the pre-trained neural network model, the neural network model needs to be trained. For example, an infrared grayscale image around the refining furnace can be collected when a flue gas leak occurs in the refining furnace, and then the corresponding target infrared grayscale image can be determined. The area with flue gas leakage in the target infrared grayscale image is marked as 0, and the area without smoke leakage is marked as 1. The marked target infrared grayscale image is used as a training sample set to train the neural network model.

[0091] Specifically, the training content is: first, the training sample set is divided into a training set and a test set according to a certain ratio, and then the training set is used to train the neural network model. During the training process, the model is evaluated by the test set, and the minimization loss function of the model is adjusted. When the minimization loss function of the model tends to converge, a trained neural network model is obtained, and then the target infrared grayscale image is input into the trained neural network model for flue gas leakage detection, thereby realizing accurate detection of flue gas leakage of the refining furnace. This embodiment does not specifically limit the division ratio of the training set and the sample set.

[0092] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for detecting a refining furnace flue gas leakage, characterized in that: include: Acquire a plurality of reference images from a plurality of consecutive frames of infrared grayscale images acquired around the refining furnace; Performing region growing with each pixel in each of the reference images as the center point to obtain the growth region of each pixel, taking the growth regions with overlapping parts as a type of growth region, screening a number of pixels in each type of growth region, where the ratio between the number of the growth regions and the total number of the growth regions in each type is greater than a preset value, to obtain the temperature change region block of each type in each of the reference images; Determine the matching area blocks of any category of variable temperature area blocks in each of the reference images, and determine the score of the matching area block belonging to smoke according to the mean of the deviation between the diffusion direction of the matching area block and the wind direction at the acquisition time of the corresponding reference image, wherein the score is negatively correlated with the mean, and the diffusion direction is the direction of the line connecting the two pixels with the farthest distance in the corresponding matching area block; The grayscale values ​​of the pixels in the matching area block are updated, and the updated grayscale values ​​are positively correlated with the score, so as to obtain a target infrared grayscale image, and input the target infrared grayscale image into a pre-trained neural network model to detect smoke leakage.

2. A refinery flue gas leakage detection method according to claim 1, characterized in that: The method for obtaining the reference image comprises: Taking the grayscale value of the pixel point at the same pixel position in each of the infrared grayscale images as the ordinate and the frame number of each of the infrared grayscale images as the abscissa, a grayscale change scatter plot is constructed; Obtaining the mean of the absolute values ​​of the slopes of the lines connecting adjacent points in the grayscale change scatter plot, the variance of the ordinates of each point, and the difference between the maximum ordinate value and the minimum ordinate value, and taking the product of the mean, the variance, and the difference as the degree of temperature change at the corresponding pixel position; The pixel points at the pixel positions whose corresponding temperature change degree is greater than or equal to the preset threshold are taken as reference pixels, and the grayscale values ​​of each reference pixel are clustered to obtain two grayscale categories of the reference pixel points. The infrared grayscale image corresponding to the category with the larger grayscale mean is taken as the reference image.

3. A refinery flue gas leakage detection method according to claim 1, characterized in that: The score of the matching area block belonging to the smoke satisfies the following relationship: ; In the formula, The score of the matching area block of any category of temperature-varying area block belonging to flue gas; The number of matching area blocks of the variable temperature area block of this category; This is the first temperature change area block of this category. The diffusion direction of the matching area blocks; The wind direction at the time of acquisition of the reference image corresponding to the matching area block; for the summation symbol; is the absolute value symbol; The natural constant An exponential function with base .

4. A refinery flue gas leakage detection method according to claim 1, characterized in that: The method for acquiring the target infrared grayscale image comprises: Calculate the gamma coefficient of the matching area block , is the score of the matching area block belonging to smoke, The maximum value of the scores of the matching area blocks corresponding to all categories of temperature-changing area blocks belonging to flue gas; The gamma coefficient is used to update the grayscale value of the pixel points in the matching area block to obtain a target infrared grayscale image, and the updated grayscale value is negatively correlated with the gamma coefficient.

5. A refinery flue gas leakage detection method according to claim 4, characterized in that: The grayscale value of the pixel in the matching area block is updated by using the gamma coefficient to satisfy the following relationship: ; In the formula, For the The first matching area block The updated gray value of each pixel; For the The first matching area block The gray value of each pixel before updating; For the The pixel value of the reference image corresponding to the matching area block; For the Gamma coefficient of the matching area block.

6. A refinery flue gas leakage detection method according to claim 1, characterized in that: The method for obtaining the matching area block includes: Arrange the reference images in order of frame numbers, take any variable temperature region block in the current frame reference image as a target region block, obtain the variable temperature region block in the next frame reference image that has the greatest matching degree with the target region block, and obtain the matching region block in the next frame reference image, wherein the matching degree reflects the difference between the area and position of the variable temperature region block in the next frame reference image and the target region block; The matching region block in the next frame of the reference image is used as a new target region block for matching, and so on, to obtain the matching region block of the target region block in each frame of the reference image.

7. A refinery flue gas leakage detection method according to claim 6, characterized in that: The matching degree satisfies the following relationship: ; In the formula, For the In the frame reference image, The matching degree between the temperature-varying area block and the target area block; For natural numbers An exponential function with base ; , are the area of ​​the target block and The area of ​​the temperature-changing region block; , are the horizontal coordinates of the central pixel of the target area block and the The horizontal coordinate of the central pixel point of the temperature-varying area block; , are the ordinates of the center pixel of the target area block, and The vertical coordinate of the central pixel of the temperature-varying area block; is the absolute value symbol.

8. A refinery flue gas leakage detection method according to claim 1, characterized in that: The method for acquiring the growth area of ​​each pixel point includes: Taking any pixel point in the reference image as a target pixel point, determining all target neighborhood pixel points within a neighborhood of the target pixel point whose grayscale value difference with the target pixel point is less than a first preset value, to form an initial growth area of ​​the target pixel point; Acquire all pixel points within the neighborhood of each target neighborhood pixel point that meet the set conditions and are within the preset growth range, obtain all growth pixel points in the initial growth area, and constitute the growth area of ​​the target pixel point; wherein the set conditions are: the difference in grayscale value with the target pixel point is less than a first preset value, and the difference in grayscale value with the corresponding target neighborhood pixel point is less than a second preset value, and the first preset value is greater than the second preset value.

Citation Information

Patent Citations

  • Adaptive contrast enhancement method for infrared image

    CN115908195A

  • VOCs gas leakage detection method and system based on optical gas imaging

    CN112781791A

  • Pipeline leakage detection method based on thermal infrared image

    CN116823839A

  • Online fault diagnosis system based on infrared image

    CN117392130A

  • Infrared imaging systems and methods for gas leak detection

    IN202017049878A