On-line monitoring and analysis method of thermal efficiency of industrial furnace based on infrared image

Abstract

The invention relates to the field of image processing technology, and in particular to an online monitoring and analysis method for the thermal efficiency of an industrial furnace based on infrared images. The method comprises: obtaining a thermal imaging image of the industrial furnace; obtaining a segmentation grayscale value of any pixel point in the thermal imaging image on three channels of R, G and B; the segmentation grayscale value is the product of the pixel value of the corresponding channel and a weight; taking the sum of the three segmentation grayscale values ​​corresponding to any pixel point as the new grayscale value of the corresponding pixel point, and forming a new grayscale image with the new grayscale values ​​of all the pixels; segmenting the new grayscale image to obtain a segmentation image to monitor and analyze the thermal efficiency of the industrial furnace; that is, the scheme of the invention can obtain an accurate segmentation image to accurately analyze the thermal efficiency of the industrial furnace.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and more specifically, to an online monitoring and analysis method for thermal efficiency of an industrial furnace based on infrared images. Background Art

[0002] Monitoring and optimizing the thermal efficiency of industrial furnaces can not only reduce energy consumption, reduce energy costs, improve the overall efficiency of the production line, and reduce equipment loss, but also save energy and reduce emissions, which is more environmentally friendly. Therefore, it is particularly important to monitor the thermal efficiency of industrial furnaces such as high-temperature hydrogen-protected induction sintering furnaces, natural gas heating furnaces, etc., in order to control the operating conditions of industrial furnaces, reduce energy waste, and improve the overall operating efficiency of industrial furnaces.

[0003] In the prior art, infrared images of industrial furnaces are usually used to obtain the temperature distribution of industrial furnaces and analyze the thermal efficiency of industrial furnaces. Before performing thermal efficiency analysis, the acquired infrared images need to be segmented to obtain effective temperature areas. The existing methods include the following two methods:

[0004] The first method is to use a region growing algorithm to segment the image to obtain the segmented region. For example, the patent document with the authorization announcement number CN116843682B and the name "Online detection and analysis system of boiler thermal efficiency using infrared thermal imager" discloses that the infrared image of the boiler is segmented by an improved region growing algorithm, and the area that can normally display the surface temperature of the boiler is selected to obtain the normal boiler area after segmentation, and then the overall temperature is obtained by the local temperature to calculate the thermal efficiency of the boiler.

[0005] Although the above scheme achieves good segmentation effect by determining the grayscale thresholds of different regions when performing boiler infrared image segmentation, it is necessary to continuously select different seed points for regional growth to obtain the final growth area; this method has low processing efficiency, and when the noise and grayscale are uneven, it may cause problems such as holes and over-segmentation.

[0006] The second method is to use the Otsu algorithm (maximum inter-class variance method) to segment the image to obtain the target area after segmentation. The core idea of ​​the traditional Otsu algorithm is to segment the image into two parts, the foreground (target) and the background, by traversing all possible thresholds, so that the intra-class variance between the two parts is minimized and the inter-class variance is maximized. At this time, the corresponding threshold is the optimal threshold, and the image is segmented using the optimal threshold. Figure 1 The thermal image of the industrial furnace is segmented to obtain Figure 2 Thermal image of industrial furnace after segmentation, which is Figure 1 Compared with the thermal imaging images in the image, there is a problem of losing important thermal distribution information.

[0007] Therefore, it is particularly important to obtain segmented images more efficiently and accurately to improve the monitoring and analysis of the thermal efficiency of industrial furnaces. Summary of the invention

[0008] The purpose of the present invention is to propose an online monitoring and analysis method for the thermal efficiency of an industrial furnace based on infrared images, so as to solve the problem that the existing technology cannot obtain segmented images more efficiently and accurately, resulting in inaccurate detection of the thermal efficiency of the industrial furnace.

[0009] The method for online monitoring and analysis of thermal efficiency of industrial furnaces based on infrared images provided by the present invention comprises:

[0010] Get thermal images of industrial furnaces;

[0011] Obtain the segmentation grayscale value of any pixel point in the thermal imaging image on the three channels of R, G, and B; the segmentation grayscale value is the product of the pixel value of the corresponding channel and the weight;

[0012] The sum of the three segmentation grayscale values ​​corresponding to any pixel is taken as the new grayscale value of the corresponding pixel, and the new grayscale values ​​of all pixels form a new grayscale image;

[0013] Segment the new grayscale image to obtain a segmentation map for monitoring and analyzing the thermal efficiency of industrial furnaces;

[0014] The weight is the normalized value of the separation importance of any channel; the separation importance for: ; D is the first degree, is the second degree of the channel graph for channel i, , are the variance and range of the areas of all connected regions in the initial segmentation graph after segmenting the channel graph for channel i, norm() is a linear normalization function, and exp() is an exponential function; channel i is any one of the three channels R, G, and B; the first degree represents the uniformity of pixel distribution in the foreground part of the extracted thermal imaging image; the second degree represents the uniformity of pixel distribution in the foreground part of the channel graph of each channel of the thermal imaging image.

[0015] The above scheme can determine the importance of separation of different channels by analyzing the foreground part in the thermal image and the foreground part of the channel image under different channels, and then determine which channel is more important for segmentation, and then assign a larger weight to the pixels on the corresponding channel to highlight the corresponding pixels, and then combine the segmentation grayscale values ​​of the pixels of all channels to obtain the new grayscale value of each pixel in the thermal image, so that the parts of the thermal image with similar colors but belonging to different temperature zones on the industrial furnace can be distinguished; that is, by using this scheme to segment the thermal image of the industrial furnace, the temperature changes in the industrial furnace can be discovered in time, and the thermal efficiency monitoring and analysis of the industrial furnace can be more accurate and sensitive.

[0016] Optionally, the first degree for: ;

[0017] in, is the sum of the gradient values ​​of the pixels in the foreground. , N are the number of non-zero pixels in the foreground and intersection graphs, respectively. is the sum of the gradient values ​​of the corresponding pixels in the thermal image for all non-zero pixels in the intersection image. is the variance of the grayscale values ​​of all non-zero pixels in the foreground, and exp( ) is an exponential function; the intersection map is the intersection of the edges of the foreground and background; the foreground and background are the regions of the thermal image extracted using the initial binary image and the reverse binary image as masks, respectively; the initial binary image is obtained by segmenting the thermal image, and the pixel values ​​of the initial binary image and the reverse binary image are both 0 or 1, and the values ​​of the pixels at the same position in the reverse binary image are opposite to those in the initial binary image.

[0018] In the above solution, by acquiring the first degree of the foreground part in the thermal imaging image, the uniformity of the pixel distribution in the foreground part can be determined.

[0019] Optionally, the second degree for: ; is the sum of the gradient values ​​of the pixels in the foreground part of the channel image when the thermal image is in channel i, , are the foreground part of the channel image and the number of non-zero pixels in the intersection image for channel i, respectively. The sum of the gradient values ​​of the corresponding pixels in the channel map for all non-zero pixels in the intersection map when it is channel i, is the variance of the grayscale values ​​of all non-zero pixels in the foreground part of the channel image at channel i, and exp() is an exponential function; the intersection map of the channel image is the intersection of the edges of the foreground and background parts of the channel image at channel i.

[0020] By analyzing the channel images on the three channels of the thermal imaging image, the uniform distribution of pixels under different channels can be obtained.

[0021] Optionally, the weight for: ; is the separation importance of thermal imaging channel i.

[0022] The above-mentioned calculation can be simplified by normalizing the separation importance.

[0023] Optionally, the initial binary image is obtained by segmenting the thermal imaging image, including:

[0024] The Otsu algorithm is used to segment the thermal image, and the pixel values ​​greater than or equal to the threshold are set to 1, and the pixel values ​​less than the threshold are set to 0 to obtain the initial binary image;

[0025] The acquisition process of the reverse binary image is as follows: the thermal imaging image is segmented by using the Otsu algorithm, pixel values ​​greater than or equal to a threshold are set to 0, and pixel values ​​less than the threshold are set to 1, to obtain the reverse binary image.

[0026] Optionally, the connected region is obtained by processing an initial segmentation map of each channel map using a connected component labeling algorithm.

[0027] Optionally, the process of segmenting the new grayscale image to obtain a segmentation image for monitoring and analyzing the thermal efficiency of the industrial furnace is as follows:

[0028] Use Otsu algorithm to segment the new grayscale image and obtain a segmentation map;

[0029] The temperature values ​​of all pixels in the segmentation map are obtained, and the thermal efficiency of the industrial furnace is calculated based on the temperature changes of the temperature values.

[0030] The above scheme can accurately obtain the thermal efficiency of the industrial furnace.

[0031] Optionally, the edges of the foreground part and the background part of the thermal imaging image and the edges of the foreground part and the background part of the channel image are obtained by using a Canny operator.

[0032] The beneficial effects of the present invention are:

[0033] The above scheme analyzes the importance of separation of thermal images in different channels to determine the importance of different channels, and then performs weighted processing on the pixel values ​​of pixels under different channels to obtain segmentation grayscale values, and then obtains new grayscale values ​​of each pixel in the thermal image, and obtains a new grayscale image of the thermal image. By segmenting it, the segmented image can be accurately obtained, and the thermal efficiency analysis of the industrial furnace can be accurately performed. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. 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:

[0035] Figure 1 A grayscale image of a thermal image of an industrial furnace is schematically shown;

[0036] Figure 2 The schematic diagram shows an initial binary image after segmentation of a thermal image of an industrial furnace;

[0037] Figure 3 A flowchart schematically shows a method for online monitoring and analysis of thermal efficiency of an industrial furnace based on infrared images;

[0038] Figure 4 The foreground portion of the thermal imaging image in this embodiment is schematically shown;

[0039] Figure 5 The background part of the thermal imaging image in this embodiment is schematically shown;

[0040] Figure 6 The schematic diagram of the edge of the foreground portion of the thermal imaging image in this embodiment is schematically shown;

[0041] Figure 7 The schematic diagram of the edge of the background part of the thermal imaging image in this embodiment is schematically shown;

[0042] Figure 8 The intersection diagram of the thermal imaging diagram in this embodiment is schematically shown;

[0043] Fig. 9 The channel diagram of the thermal imaging image in the B channel in this embodiment is schematically shown;

[0044] Fig.10 The channel diagram of the thermal imaging image in the R channel in this embodiment is schematically shown;

[0045] Fig.11 The new grayscale image of the thermal imaging image in this embodiment is schematically shown;

[0046] Fig.12 The segmentation diagram after segmentation in this embodiment is schematically shown. DETAILED DESCRIPTION

[0047] 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.

[0048] like Figure 3 As shown, the online monitoring and analysis method of thermal efficiency of an industrial furnace based on infrared images in this embodiment includes the following steps:

[0049] Step S1, obtaining a thermal image of an industrial furnace.

[0050] In this embodiment, a thermal imaging image of the industrial furnace is collected by a thermal imager installed near the industrial furnace, wherein the shooting position needs to be at an angle that can capture the complete industrial furnace as much as possible.

[0051] The thermal image is a visual representation of the temperature of various parts of the industrial furnace surface. It reflects the strength of infrared radiation from various parts of the industrial furnace surface, thereby showing the temperature difference of various parts of the industrial furnace surface.

[0052] Step S2, obtaining a new grayscale image of the thermal image. Specifically, in this embodiment, by obtaining the segmentation grayscale values ​​of any pixel in the thermal image on the three channels of R, G, and B; taking the sum of the three segmentation grayscale values ​​corresponding to any pixel as the new grayscale value of the corresponding pixel, the new grayscale values ​​of all pixels constitute a new grayscale image.

[0053] In this embodiment, the process of obtaining the new grayscale value includes steps S21 to S24, which are specifically as follows:

[0054] Step S21, acquiring the foreground part and the background part of the thermal imaging image, and obtaining an intersection image of the edge of the foreground part and the edge of the background part.

[0055] The acquisition process of the foreground part is as follows: the acquired thermal image of the industrial furnace is initially segmented to obtain an initial binary image, wherein the pixel values ​​in the initial binary image are 0 or 1 (pixel values ​​greater than or equal to the threshold are set to 1, and pixel values ​​less than the threshold are set to 0); the initial binary image is used as a mask to extract the foreground part of the thermal image; illustratively, the initial segmentation image is used as a mask, the pixel values ​​of the pixels at the same position in the thermal image as the pixels with a pixel value of 0 in the mask are set to 0, and the pixel values ​​of the pixels at other positions remain unchanged, thereby obtaining the foreground part of the thermal image, such as Figure 4 shown.

[0056] The foreground portion mentioned above is actually the target area in the thermal image.

[0057] The above initial segmentation method uses the Otsu algorithm; of course, the histogram method can also be used.

[0058] Before the initial segmentation of the thermal image, the thermal image of the industrial furnace is grayed to obtain a gray image.

[0059] The process of obtaining the background part is as follows: using the reverse binary image as a mask to obtain the background part of the thermal image, such as Figure 5 As shown; the values ​​of the pixels in the reverse binary image are also 0 or 1, but the values ​​of the pixels at the same position in the reverse binary image are opposite to those in the initial binary image.

[0060] The process of obtaining the reverse binary image is as follows: the thermal image is segmented using the Otsu algorithm, pixel values ​​greater than or equal to the threshold are set to 0, and pixel values ​​less than the threshold are set to 1, to obtain the reverse binary image.

[0061] As another implementation, the pixel values ​​of the pixels with pixel values ​​of 1 in the initial binary image may be set to 0, and the pixel values ​​of the pixels with pixel values ​​of 0 may be set to 1, so as to obtain an inverse binary image.

[0062] The above threshold is the optimal threshold obtained by the Otsu algorithm, and since it is a prior art, it will not be described in detail.

[0063] The process of obtaining the above intersection graph is: use the Canny operator to extract the edge of the foreground part and the edge of the background part to obtain the foreground edge (see Figure 6 ) and the background edge (see Figure 7 ), obtain the intersection graph generated by the intersection between the foreground edge and the background edge, such as Figure 8 As shown, in the intersection map, the pixel values ​​of the pixels that overlap the edge of the foreground part and the edge of the background part are set to 1, and the pixel values ​​of the pixels at other positions are set to 0.

[0064] Step S22, calculating the first degree of the foreground part.

[0065] Due to the inherent defects of the Otsu algorithm, the segmentation effect of the thermal image of the industrial furnace will be not detailed, which will cause the parts of the thermal image with different thermal distribution information to be segmented into the same category, affecting the monitoring and analysis of the thermal efficiency of the industrial furnace. Therefore, in this embodiment, it is necessary to calculate the first degree of the foreground part of the thermal image.

[0066] Among them, the specific process of obtaining the first degree is:

[0067] The image gradient of the foreground part of the thermal image is calculated by the Sobel operator;

[0068] The pixel points whose pixel values ​​are not equal to 0 in the image are regarded as non-zero pixel points, and the variance of the grayscale values ​​of all non-zero pixel points in the foreground part of the thermal image is calculated;

[0069] The first degree of the foreground part of the thermal image is calculated by image gradient and variance.

[0070] Specifically, the first degree is: ; is the first degree of the foreground part, is the sum of the gradient values ​​of all pixels in the foreground. is the number of non-zero pixels in the foreground. is the sum of the gradient values ​​of the corresponding pixels in the thermal image for all non-zero pixels in the intersection image. is the variance of the grayscale values ​​of all non-zero pixels in the foreground, N is the number of non-zero pixels in the intersection map, and exp() is an exponential function with the natural constant e as the base.

[0071] Since the intersection map is composed of the intersection of the edge of the foreground and the edge of the background, and the position of the non-zero pixel point in the intersection map corresponds to the contour line between the foreground and the background in the thermal image, is the number of all non-contour pixels in the foreground part of the thermal image, and is the sum of the gradient values ​​of all non-contour pixels in the foreground of the thermal image. Since the contour line of the foreground of the thermal image has a higher gradient, and the high-gradient pixels on the contour line have no meaning for the first degree calculation of the foreground of the thermal image, and Perform a first degree calculation, It is the average gradient value of all non-contour pixels in the foreground part of the thermal image. This value represents the overall level of the gradient value of the pixels in the foreground part of the thermal image. The larger the value, the more uneven the color distribution of the pixels in the foreground part of the thermal image, and the lower the first degree of the foreground part of the thermal image. The smaller the value, the more uniform the color distribution of the pixels in the foreground part of the thermal image, and the higher the first degree of the foreground part of the thermal image.

[0072] In addition, for the convenience of calculation, the Normalize.

[0073] The above variance represents the color distribution of pixels in the foreground part of the thermal image. The larger the variance, the more uneven the color distribution of pixels in the foreground part of the thermal image, and the lower the first degree of the foreground part of the thermal image; the smaller the variance, the more uniform the color distribution of pixels in the foreground part of the thermal image, and the higher the first degree of the foreground part of the thermal image.

[0074] At the same time, in order to facilitate calculation, the Normalize.

[0075] Step S23, calculating the second degree of the thermal image in any one of the three channels R, G, and B; and determining the separation importance of any one of the channels according to the first degree and the second degree of the any one of the channels.

[0076] The thermal image is composed of the channel images of the three channels R, G, and B. Different colors in the thermal image are formed by different pixel values ​​on the three channels. However, it should be noted that when the thermal image is segmented, not all channels have an effect on the segmentation of the thermal image. For example, Fig. 9 and Fig.10 As shown, Fig. 9 This is the channel image of the B channel of the thermal image. Fig.10 This is the channel diagram of the R channel of the thermal image. It can be seen that in this scene, the pixel value of the red channel cannot play a positive role in the detailed segmentation of the image. Therefore, it is necessary to evaluate the uniformity of the pixel distribution of the three channels (i.e., the second degree) to determine the importance of each channel in the segmentation of the thermal image.

[0077] It should be noted that the pixel value of the red channel in this example cannot play a positive role in the detailed segmentation of the image. In other scenarios, the pixel value of the red channel may play a positive role in the segmentation of the image, which is analyzed specifically according to the specific scenario.

[0078] The calculation method of the second degree in this embodiment is the same as that of the first degree, and the only difference is that the object of calculation is different, that is, the object of the second degree is the channel map of any channel among the three channels of R, G, and B of the thermal imaging image; the second degree represents the uniformity of pixel distribution in the foreground part of any channel map.

[0079] Specifically, the second degree acquisition process is:

[0080] First, the channel map of each channel of the thermal image is obtained, and the channel maps of the three channels are segmented respectively to obtain the initial segmentation map of each channel map (the initial segmentation map is a binary map) and the reverse segmentation map. The above initial segmentation map is obtained by the Otsu algorithm.

[0081] Secondly, based on the initial segmentation map and the reverse segmentation map, the foreground and background parts of each channel map are extracted; and then the intersection map of each channel map is obtained;

[0082] Then, the second degree of each channel map is calculated.

[0083] Taking the R channel as an example, calculate the second degree of the foreground part of the channel image of the thermal image in the R channel for:

[0084] ; is the sum of the gradient values ​​of the pixels in the foreground part of the channel image when the thermal image is in channel R. , They are the foreground part of the channel image and the number of non-zero pixels in the intersection image when channel R is used. When channel R is the channel graph, the sum of the gradient values ​​of all non-zero pixels in the intersection graph of the channel graph in the corresponding channel graph. When it is channel R, it is the variance of the grayscale values ​​of all non-zero pixels in the foreground part of the channel image, and exp() is an exponential function.

[0085] Furthermore, in this embodiment, after obtaining the second degree of each channel, the separation importance of each channel is also calculated. The specific process is:

[0086] The segmentation graph of each channel is processed by the connected component labeling algorithm to obtain the number of connected regions and the area of ​​each connected region (the number of pixels contained in the region) in each channel segmentation graph; the separation importance of the corresponding channel is obtained according to the second degree, the first degree, the number of connected regions and the area of ​​each connected region. Since the connected component labeling algorithm is a prior art, it will not be described here.

[0087] Specifically, the separation importance is: ;

[0088] in, is the separation importance of the channel graph for channel i, For the first degree, is the second degree of the channel graph for channel i, is the variance of the area of ​​all connected regions in the initial segmentation map of the thermal image under channel i, is the range of the areas of all connected regions in the initial segmentation map of the channel map of the thermal image under channel i, norm( ) is a linear normalization function, exp( ) is an exponential function with the natural constant e as the base; channel i is any one of the three channels R, G, and B.

[0089] is the possibility that there are incompletely segmented connected regions in the initial segmentation map under channel i, The larger the value of is, the larger the difference between the areas of the connected regions in the initial segmentation map under channel i is. The more likely it is that there are incompletely segmented connected regions in the initial segmentation map under channel i, and the less important the channel is for the segmentation of the thermal image. The less important the separation of the channel map is for channel i. The smaller the value of is, the smaller the difference between the areas of the connected regions in the initial segmentation map under channel i is, and the connected regions in the initial segmentation map under channel i are more likely to be segmented completely. The more important the channel is to the segmentation of the thermal imaging image, the greater the importance of the channel map separation at channel i.

[0090] is the ratio of the first degree to the second degree of the channel graph for channel i, The smaller the The larger it is, the worse the segmentation effect of the thermal image through the channel map of channel i is; The larger the The smaller the value is, the channel image at channel i is conducive to the segmentation of the thermal image. right Make corrections, The smaller it is (the more likely it is to be less than 1), right It plays an upward correction role, making the separation importance of the i-th channel greater. The larger it is (the more likely it is to be greater than 1), then right It plays a downward correction role, making the separation importance of the i-th channel smaller.

[0091] Step S24, using the normalized value of the separation importance as the weight of the pixel value of each channel image, obtaining the segmentation grayscale value of the pixel point of each channel, and taking the sum of the segmentation grayscale values ​​of any pixel point in the thermal imaging image on the three channels of R, G, and B as the new grayscale value.

[0092] Among them, the segmentation gray value is the product of the weight of the corresponding channel and the pixel value on the corresponding channel. Then the sum of the segmentation gray values ​​at the same position on the three channels is used as the new gray value of the corresponding position in the thermal image.

[0093] In this embodiment, the weight of each channel of the thermal imaging image is obtained by normalizing the separation importance of each channel. , specifically:

[0094] ; is the separation importance of the channel graph when channel i. The larger the value, the more important the channel i is for the segmentation of thermal images. The smaller it is, the smaller the role of channel i in the segmentation of the thermal image, and the smaller the weight of the channel.

[0095] As another implementation, a maximum and minimum value normalization method may also be adopted.

[0096] Since the pixel value of each channel of the thermal imaging image has different effects on the expression of color in the thermal imaging image, the pixel value of each channel of the thermal imaging image will also play different roles in the segmentation of the thermal imaging image. Therefore, in this embodiment, the segmentation grayscale value under different channels is determined according to the importance of separation of each channel, and then the new grayscale value of each pixel point in the thermal imaging image is determined to facilitate subsequent segmentation.

[0097] In this embodiment, all new grayscale values ​​form a new grayscale image, such as Fig.11 shown.

[0098] Step S3, segmenting the new grayscale image to obtain a segmentation image for monitoring and analyzing the thermal efficiency of the industrial furnace.

[0099] In this embodiment, the Otsu algorithm is used to segment the new grayscale image to obtain a segmentation image, such as Fig.12 shown.

[0100] In this embodiment, the infrared thermal imager can be used to directly obtain the temperature of each pixel in the segmentation image. The temperature change, maximum temperature and other factors can be calculated based on the surface temperature of the industrial furnace within a period of time, thereby calculating the thermal efficiency of the industrial furnace.

[0101] As another implementation method, the segmentation map and the thermal imaging image of the industrial furnace are input into the FLIR tools software to generate a temperature value file of all pixels of the grayscale image of the industrial furnace, which is used to measure the surface temperature of the boiler; the local temperature average is used as the overall temperature average to calculate the thermal effect, and the temperature of the segmented affected area is inferred by linear interpolation to perform the final thermal effect calculation. Since the thermal effect calculation is a well-known technology, it will not be described in detail here.

[0102] Furthermore, after the segmentation map is obtained, the segmentation map may contain multiple segmentation areas, so the multiple segmentation areas can be screened, that is, the segmentation area containing the number of pixels greater than the preset number threshold is taken as the target part of the industrial furnace; the thermal efficiency of the target part of the industrial furnace is monitored. The preset number threshold is not less than one third of the pixels in the thermal imaging image, and of course the preset number threshold can also be set according to actual conditions.

[0103] The solution of the present invention can obtain more detailed segmentation results by performing image processing on the thermal imaging image of the industrial furnace, so that different areas with similar temperatures in the thermal imaging image can be distinguished, the distribution of different temperature zones in the industrial furnace can be discovered in time, and the thermal efficiency of the industrial furnace can be monitored and analyzed for the distinguished different areas.

[0104] In the description of this specification, “plurality” means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0105] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. The online monitoring and analysis method of thermal efficiency of industrial furnaces based on infrared images is characterized by: include: Get thermal images of industrial furnaces; Obtain the segmentation grayscale value of any pixel point in the thermal imaging image on the three channels of R, G, and B; the segmentation grayscale value is the product of the pixel value of the corresponding channel and the weight; The sum of the three segmentation grayscale values ​​corresponding to any pixel is taken as the new grayscale value of the corresponding pixel, and the new grayscale values ​​of all pixels form a new grayscale image; Segment the new grayscale image to obtain a segmentation map for monitoring and analyzing the thermal efficiency of industrial furnaces; The weight is the normalized value of the separation importance of any channel; the separation importance for: ; D is the first degree, is the second degree of the channel graph for channel i, , are the variance and range of the areas of all connected regions in the initial segmentation graph after segmenting the channel graph for channel i, norm() is a linear normalization function, and exp() is an exponential function; channel i is any one of the three channels R, G, and B; the first degree represents the uniformity of pixel distribution in the foreground part of the extracted thermal imaging image; the second degree represents the uniformity of pixel distribution in the foreground part of the channel graph of each channel of the thermal imaging image; The process of obtaining the first degree is as follows: the image gradient of the foreground part of the thermal image is calculated by the Sobel operator; the pixel points whose pixel values ​​are not equal to 0 in the image are taken as non-zero pixel points, and the variance of the grayscale values ​​of all non-zero pixel points in the foreground part of the thermal image is calculated; the first degree of the foreground part of the thermal image is calculated by the image gradient and variance; The process of obtaining the second degree is as follows: obtaining the channel map of each channel of the thermal imaging image, segmenting the channel maps of the three channels respectively, and obtaining the initial segmentation map and the reverse segmentation map of each channel map; the initial segmentation map is obtained by the Otsu algorithm; based on the initial segmentation map and the reverse segmentation map, the foreground part and the background part of each channel map are extracted; and then the intersection map of each channel map is obtained; then, the second degree of each channel map is calculated, and the calculation method of the second degree is the same as that of the first degree.

2. The method for online monitoring and analysis of thermal efficiency of industrial furnaces based on infrared images according to claim 1 is characterized in that: The first degree for: ; in, is the sum of the gradient values ​​of the pixels in the foreground. , N are the number of non-zero pixels in the foreground and intersection graphs, respectively. is the sum of the gradient values ​​of the corresponding pixels in the thermal image for all non-zero pixels in the intersection image. is the variance of the grayscale values ​​of all non-zero pixels in the foreground, and exp( ) is an exponential function; the intersection map is the intersection of the edges of the foreground and background; the foreground and background are the regions of the thermal image extracted using the initial binary image and the reverse binary image as masks, respectively; the initial binary image is obtained by segmenting the thermal image, and the pixel values ​​of the initial binary image and the reverse binary image are both 0 or 1, and the values ​​of the pixels at the same position in the reverse binary image are opposite to those in the initial binary image.

3. The method for online monitoring and analysis of thermal efficiency of industrial furnaces based on infrared images according to claim 1 is characterized in that: The second degree for: ; is the sum of the gradient values ​​of the pixels in the foreground part of the channel image when channel i is , are the foreground part of the channel image and the number of non-zero pixels in the intersection image for channel i, respectively. The sum of the gradient values ​​of the corresponding pixels in the channel map for all non-zero pixels in the intersection map when it is channel i, is the variance of the grayscale values ​​of all non-zero pixels in the foreground part of the channel image at channel i, and exp() is an exponential function; the intersection map of the channel image is the intersection of the edges of the foreground and background parts of the channel image at channel i.

4. The method for online monitoring and analysis of thermal efficiency of industrial furnaces based on infrared images according to claim 1 is characterized in that: The weight for: ; is the separation importance of thermal imaging channel i.

5. The method for online monitoring and analysis of thermal efficiency of industrial furnaces based on infrared images according to claim 2 is characterized in that: The initial binary image is obtained by segmenting the thermal imaging image, including: The Otsu algorithm is used to segment the thermal image, and the pixel values ​​greater than or equal to the threshold are set to 1, and the pixel values ​​less than the threshold are set to 0 to obtain the initial binary image; The acquisition process of the reverse binary image is as follows: the thermal imaging image is segmented by using the Otsu algorithm, pixel values ​​greater than or equal to a threshold are set to 0, and pixel values ​​less than the threshold are set to 1, to obtain the reverse binary image.

6. The method for online monitoring and analysis of thermal efficiency of industrial furnaces based on infrared images according to claim 1 is characterized in that: The connected regions are obtained by processing the initial segmentation graphs of each channel graph using a connected component labeling algorithm.

7. The method for online monitoring and analysis of thermal efficiency of industrial furnaces based on infrared images according to claim 1 is characterized in that: The process of segmenting the new grayscale image to obtain a segmentation image for monitoring and analyzing the thermal efficiency of the industrial furnace is as follows: Use Otsu algorithm to segment the new grayscale image and obtain a segmentation map; The temperature values ​​of all pixels in the segmentation map are obtained, and the thermal efficiency of the industrial furnace is calculated based on the temperature changes of the temperature values.

8. The method for online monitoring and analysis of thermal efficiency of industrial furnaces based on infrared images according to claim 2 is characterized in that: The edges of the foreground and background of the thermal imaging image and the edges of the foreground and background of the channel image are obtained by using the Canny operator.

Images