Steam Boiler Gas Leak Detection Method Based on Image Processing

Abstract

The present invention relates to the field of image data processing, and particularly to a method for detecting gas leakage in a steam boiler based on image processing. The method includes: acquiring a series of consecutive grayscale images of the position to be measured in the steam boiler; obtaining the connected regions in each grayscale image; calculating the irregularity degree of the connected regions according to the circularity and area differences of the connected regions in each of the series of consecutive grayscale images. When the irregularity degree of the connected regions is greater than a threshold, there is gas leakage at the position to be measured. The irregularity degree of the connected regions is inversely correlated with the circularity of the connected regions in each grayscale image, and is positively correlated with the difference between the area of the connected regions in each grayscale image and the average value of the areas of the connected regions in the series of consecutive grayscale images. The present invention effectively solves the problem in the prior art that gas leakage in a steam boiler cannot be accurately detected.

Description

Technical Field

[0001] The present invention relates to the field of image data processing. More specifically, the present invention relates to a method for detecting gas leakage of a steam boiler based on image processing. Background Art

[0002] In the detection of gas leakage of a steam boiler, the image processing method can capture the operating state of the boiler through a monitoring device, and analyze the possible leakage signs in the image in real time, improving the detection efficiency and accuracy. The prior art judges whether leakage occurs by performing image enhancement on the leakage gas image to extract relevant gas features.

[0003] However, since the gas concentration is low when the steam boiler just starts to leak and the contrast of the features of the gas on images of different scales is different, the traditional CLAHE algorithm performs image enhancement based on a fixed local threshold and a single scale, and cannot accurately extract the edges and connected domains of the gas, resulting in untimely leakage identification or misidentification.

[0004] Currently, in the Chinese patent document with the authorization announcement number CN116563283B and the name of "Method and Detection Device for Detecting Gas Leakage of Steam Boiler Based on Image Processing", the specific method is as follows: determining a first abnormal area of each infrared image in the infrared image set, and determining a second abnormal area of each visible light image in the visible light image set; calculating the similarity of the two abnormal areas, and determining the human interference area in the first abnormal area based on the similarity; and removing the human interference area in the first abnormal area to obtain a third abnormal area; then determining whether the steam boiler leaks.

[0005] The above uses infrared images and visible light images to detect the gas leakage situation of the steam boiler. However, the infrared images are greatly affected by temperature, and the temperature near the steam boiler is generally relatively high. The collected infrared images may have poor accuracy, easily causing false detection and resulting in the inability to accurately detect the gas leakage of the steam boiler. Summary of the Invention

[0006] To solve the problem of the inability to accurately detect the gas leakage of the steam boiler, the present invention provides the following solutions.

[0007] The present invention provides a method for detecting gas leakage of a steam boiler based on image processing, including: obtaining a plurality of consecutive grayscale images of the position to be measured of the steam boiler; presetting a plurality of scales, each grayscale image is segmented into sub-regions of the same area according to the scale, and the area of the sub-regions corresponding to each scale is different, and using the CLAHE algorithm to obtain a plurality of enhanced images corresponding to each grayscale image at multiple scales; when the probability that a pixel point in the grayscale image belongs to an edge pixel point exceeds a preset probability, the pixel point is an edge pixel point, and the connected domains in each grayscale image are obtained according to the edge pixel points; the probability that a pixel point in the grayscale image belongs to an edge pixel point , , where is normalization, is the number of scale types, is the difference in the noise level change of the enhanced image corresponding to scale k, is the gradient magnitude of pixel point z in the enhanced image corresponding to scale k, is the maximum value of the gradient magnitudes of all pixel points in the enhanced image corresponding to scale k; The irregularity degree of the connected region is calculated based on the circularity and area difference of the connected regions in each frame of grayscale image. The irregularity degree of the connected region is inversely correlated with the circularity of the connected region in each frame of grayscale image and positively correlated with the difference between the area of the connected region in each frame of grayscale image and the average area of the connected regions in consecutive frames of grayscale images; When the irregularity degree of the connected region is greater than the threshold, there is a gas leak at the position to be measured.

[0008] By processing the image at different scales, different details can be enhanced, avoiding noise interference at a certain scale, so as to more accurately extract edge information. Through the analysis of gradient magnitude and noise change, the probability that each pixel point belongs to the edge can be calculated more precisely, thereby improving the accuracy of edge detection. Using geometric features such as circularity and area difference, the connected region features of whether there is a gas leak can be effectively distinguished. This method can judge in real time whether there is a gas leak at the position of the steam boiler according to consecutive frame images, improving the detection accuracy of gas leaks.

[0009] Preferably, the irregularity degree of the connected region includes: , where represents the irregularity degree of the connected region, represents the total number of frames of consecutive frame grayscale images, represents the circularity of the connected region on the th frame of grayscale image, represents the area of the connected region on the th frame of grayscale image,

[0010] When there is a gas leak in the steam boiler, the connected regions formed by the diffusion of smoke or gas are often irregular, and the area will fluctuate greatly over time. In normal pictures, the edge connected regions are more regular and the area will not fluctuate greatly. Combining circularity and area changes, it is possible to more accurately judge whether there is a gas leak.

[0011] Preferably, obtaining multiple enhanced images corresponding to each frame of grayscale image at multiple scales by using the CLAHE algorithm further includes: for each sub-region obtained from each frame of grayscale image at each scale, setting a corresponding adaptive contrast limiting threshold respectively, where the adaptive contrast limiting threshold is positively correlated with the noise level of the sub-region and negatively correlated with the sum of the noise levels of all sub-regions.

[0012] Since the contrast limiting threshold adapts to the local noise level and changes, it can effectively reduce the over-enhancement of high-noise regions and prevent the amplification of noise.

[0013] Preferably, the adaptive contrast limiting threshold is specifically: , , where is the adaptive contrast limiting threshold of the i-th sub-region, is the initial contrast limiting threshold, is the adjustment parameter of the i-th sub-region, and are the noise levels of the i-th and j-th sub-regions, and N is the number of sub-regions.

[0014] Preferably, the noise level of the sub-region includes: , where represents the noise level of the t-th sub-region, represents the number of pixel points in the sub-region, represents the grayscale value of the w-th pixel point in the sub-region, represents the average grayscale value of all pixel points in the t-th sub-region.

[0015] Preferably, obtaining the connected regions of each frame of grayscale image includes: for each frame of grayscale image, calculating the difference in noise level change of the enhanced image at each scale, , where is the difference in noise level change of the enhanced image corresponding to scale k, and are the average noise levels of all sub-regions of the enhanced images corresponding to scales k and h, and are the average noise levels of all sub-regions after dividing the grayscale image into sub-regions according to scales k and h, and M is the number of scale types; using the sobel operator to calculate the gradient magnitude of each pixel point in the enhanced images at different scales, and calculating the probability that each pixel point in the grayscale image belongs to an edge pixel point in combination with the difference in noise level change of the enhanced image at each scale.

[0016] Combining the gradient information at multiple scales with the difference in noise level change can more accurately extract edges and reduce false detection and missed detection.

[0017] Preferably, obtaining the connected regions in each frame of grayscale image based on the edge pixel points includes:

[0018] Obtaining the connected regions by using depth - first search or breadth - first search.

[0019] Preferably, obtaining the continuous multi - frame grayscale images of the position to be measured of the steam boiler includes: using a high - definition camera to capture continuous multi - frame visible - light images of the position to be measured of the steam boiler, and converting the continuous multi - frame visible - light images into grayscale images.

[0020] Preferably, when there is a gas leak at the position to be measured, an audible and visual alarm is given.

[0021] When a gas leak occurs, it can provide timely and effective feedback in a short time, enabling the staff to react quickly and take measures to prevent the accident from expanding.

[0022] Preferably, the threshold is 0.6.

[0023] The beneficial effects of the present invention are as follows: By processing images at different scales, different details can be enhanced, avoiding noise interference at a certain scale, so as to more accurately extract edge information. Through gradient magnitude and noise change analysis, the probability that each pixel point belongs to the edge can be calculated more precisely, thereby improving the accuracy of edge detection. Using geometric features such as circularity and area difference, the connected - region features of whether there is a gas leak can be effectively distinguished. This method can judge in real time whether there is a gas leak at the position of the steam boiler according to continuous - frame images, improving the detection accuracy of gas leaks. Description of the Drawings

[0024] Figure 1 is a flowchart of a method for detecting gas leaks in a steam boiler based on image processing provided by an embodiment of the present invention. Detailed Embodiments

[0025] The following will describe in detail the specific embodiments of the present invention with reference to the drawings.

[0026] Figure 1 is a flowchart of a method for detecting gas leaks in a steam boiler based on image processing provided by an embodiment of the present invention, including the following steps:

[0027] S101. Obtain continuous multi - frame grayscale images of the position to be measured of the steam boiler.

[0028] In some embodiments, obtaining the continuous multi - frame grayscale images of the position to be measured of the steam boiler includes: using a high - definition camera to capture continuous multi - frame visible - light images of the position to be measured of the steam boiler, and converting the continuous multi - frame visible - light images into grayscale images. The method of converting visible - light images into grayscale images is a well - known technology and will not be elaborated here.

[0029] In the above embodiment, the visible light image of the position to be measured is directly captured. In another embodiment, the visible light image of the entire steam boiler can also be captured, and the partial image containing the position to be measured is intercepted from the entire image.

[0030] S102. Obtain the connected components in each grayscale image.

[0031] The method for obtaining the connected components in each grayscale image specifically includes: presetting multiple scales, dividing each grayscale image into sub-regions of the same area according to the scales, and the area of the sub-regions corresponding to each scale is different. Using the CLAHE algorithm to obtain multiple enhanced images corresponding to each grayscale image at multiple scales; when the probability that a pixel point in the grayscale image belongs to an edge pixel point exceeds a preset probability, the pixel point is an edge pixel point, and the connected components in each grayscale image are obtained according to the edge pixel points.

[0032] In one embodiment, in an embodiment, the gases leaked in the steam boiler mainly include water vapor, carbon monoxide, carbon dioxide, etc. Since the concentration of these gases is relatively low when the steam boiler starts to leak, the color contrast in the grayscale image is relatively low, and the edges of the gases are relatively blurred, resulting in inaccurate extraction of edge points, untimely leakage recognition or misjudgment of leakage when the color contrast in the grayscale image is relatively low and the steam edge is relatively blurred. The traditional CLAHE algorithm usually applies a fixed contrast limiting threshold to each sub-region, which may cause loss of details in some regions or over-enhancement in some regions with more noise. Therefore, the CLAHE algorithm can be improved, specifically including: for the sub-regions obtained from each grayscale image at each scale, an adaptive contrast limiting threshold is set for each sub-region, and the adaptive contrast limiting threshold is positively correlated with the noise level of the sub-region and negatively correlated with the sum of the noise levels of all sub-regions. In this embodiment, by adjusting the adaptive contrast limiting threshold according to the noise level, the contrast can be enhanced in the regions with less noise, and over-enhancement in the regions with more noise can be avoided, thereby improving the quality of the enhanced image.

[0033] Specifically, the adaptive contrast limiting threshold , , , where is the initial contrast limiting threshold, is the adjustment parameter of the i-th sub-region, and are the noise levels of the i-th and j-th sub-regions, N is the number of sub-regions, The magnitude of represents the The intensity of the noise level of the sub-region. In some embodiments, the noise level of the sub-region includes: , where represents the noise level of the t-th sub-region, represents the number of pixel points in the sub-region, represents the gray value of the w-th pixel point in the sub-region, represents the average gray value of all pixel points in the t-th sub-region.

[0034] After obtaining multiple enhanced images of each frame of grayscale image, it is necessary to comprehensively analyze these enhanced images to obtain the edge pixel points in the grayscale image of this frame, so as to obtain the connected components of the grayscale image. Obtaining the connected components of each frame of grayscale image specifically includes: for each frame of grayscale image, calculating the difference in the change of the noise level of the enhanced image at each scale, , where is the difference in the change of the noise level of the enhanced image corresponding to scale k, and are the average noise levels of all sub-regions of the enhanced images corresponding to scales k and h, and are the average noise levels of all sub-regions after dividing the grayscale image into sub-regions according to scales k and h, M is the number of scale types; using the sobel operator to calculate the gradient magnitude of each pixel point in the enhanced images at different scales, and calculating the probability that each pixel point in the grayscale image belongs to an edge pixel point in combination with the difference in the change of the noise level of the enhanced image at each scale. Obtaining the connected components in each frame of grayscale image according to the edge pixel points includes: using depth-first search or breadth-first search to obtain the connected components.

[0035] The probability that a pixel point in the grayscale image belongs to an edge pixel point , , where is normalization, is the number of scale types, is the difference in the change of the noise level of the enhanced image corresponding to scale k, is the gradient magnitude of pixel point z in the enhanced image corresponding to scale k, is the maximum value of the gradient magnitudes of all pixel points in the enhanced image corresponding to scale k. Generally, when the probability is greater than 0.7, it can be considered that pixel point z in the grayscale image is an edge pixel point.

[0036] For example, three scales are preset, and each scale corresponds to a sub-region area: Scale 1, sub-region area a×a; Scale 2, sub-region area b×b; Scale 3, sub-region area c×c. Each grayscale image frame is divided into A sub-regions of a×a according to Scale 1, and the enhanced image of this grayscale image frame at Scale 1 is obtained using the CLAHE algorithm. Similarly, the enhanced images at Scale 2 and Scale 3 can be obtained. Thus, a total of three enhanced images of this grayscale image frame at three scales are obtained. According to the difference in the noise level change corresponding to these three images and the gradient magnitude of each pixel point, the probability that each pixel point in this grayscale image belongs to an edge pixel point is calculated. Those exceeding the probability threshold are edge pixel points. Then, based on the edge pixel points in this grayscale image, the connected components of this grayscale image can be obtained using depth-first search or breadth-first search. Obtaining the connected components based on the edge pixel points belongs to well-known technology and will not be elaborated here.

[0037] S103. Calculate the irregularity degree of the connected components based on the circularity and area difference of the connected components in each frame of the consecutive grayscale image frames. When the irregularity degree of the connected components is greater than the threshold, there is a gas leak at the position to be detected.

[0038] The irregularity degree of the connected components is inversely correlated with the circularity of the connected components in each frame of the grayscale image and is positively correlated with the difference between the area of the connected components in each frame of the grayscale image and the average value of the areas of the connected components in consecutive grayscale image frames.

[0039] Due to the high-temperature and high-pressure environment inside the steam boiler, water vapor evaporated at high temperature will uniformly appear around the steam boiler, and the shape of the water vapor is relatively regular. When a leak occurs, due to the large pressure difference inside and outside, the leaked gas will show an irregular and continuously changing phenomenon. The circularity can reflect whether the leaked gas is regular in shape, and the area change difference between different frames can reflect the degree of area change of the leaked gas. Therefore, the present invention calculates the irregularity degree of the connected components at the detection position based on the circularity and area difference of the connected components in consecutive frame grayscale images. The greater the difference in the area of the connected components between different frames and the lower the circularity of the connected components, the higher the irregularity degree of the connected components, and the greater the probability of a leak.

[0040] The present invention provides a formula for calculating the irregularity degree of the connected components. Specifically, , where represents the irregularity degree of the connected components, represents the total number of frames of the consecutive grayscale image frames, represents the circularity of the connected components on the th frame of the grayscale image, represents the area of the connected components on the It represents the average area of connected components on all consecutive frame grayscale images. In one embodiment, the threshold is 0.6. When the irregularity degree of the connected component is greater than 0.6, it indicates that there is a gas leak at the position to be measured. At this time, an acoustic and optical warning can be sent to the control center in a timely manner. The acoustic and optical warning can provide timely and effective warning in a short time, enabling the staff to quickly know the boiler leak and take measures to prevent the accident from expanding.

[0041] Through the preset sub-region segmentation of different scales, the present invention can capture edge features of different sizes and finally obtain more accurate connected components. The CLAHE algorithm restricts the amplitude of local histogram equalization, suppressing noise amplification while enhancing the image. Through the quantitative analysis of the irregularity degree of the connected components in the image, the morphological changes in the gas leak area can be sensitively captured. This method combines multiple technologies such as multi-scale image processing, edge detection, and connected component analysis, effectively improving the accuracy of gas leak detection.

[0042] In the description of this specification, the meanings of "a plurality of" and "several" are at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0043] Although this specification has shown and described multiple embodiments of the present invention, it is obvious to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will think of many changes, alterations, and alternative ways without departing from the spirit and idea of the present invention. It should be understood that various alternative solutions to the embodiments of the present invention described herein can be adopted in the practice of the present invention.

Claims

1. A steam boiler gas leakage detection method based on image processing, characterized in that: include: Acquire continuous multiple-frame grayscale images of the position to be tested of the steam boiler; Multiple scales are preset, and each frame of grayscale image is divided into sub-regions of the same area according to the scale. Each scale corresponds to a different sub-region area. The CLAHE algorithm is used to obtain multiple enhanced images corresponding to each frame of grayscale image at multiple scales. When the probability that a pixel in the grayscale image belongs to an edge pixel exceeds a preset probability, the pixel is an edge pixel, and a connected domain in each frame of the grayscale image is obtained based on the edge pixel; The probability that a pixel in the grayscale image belongs to an edge pixel , ,in, For normalization, is the number of scale types, is the noise level change difference of the enhanced image corresponding to scale k, is the gradient amplitude of pixel z in the enhanced image corresponding to scale k, is the maximum value of the gradient amplitude of all pixels in the enhanced image corresponding to scale k; Get the connected domain of each frame of grayscale image, including: For each frame of grayscale image, calculate the difference in noise level change of the enhanced image at each scale. ,in, is the noise level change difference of the enhanced image corresponding to scale k, and is the average noise level of all sub-regions of the enhanced image corresponding to scale k and scale h, and is the average noise level of all sub-regions of the grayscale image after the sub-regions are divided according to scale k and scale h, and M is the number of scale types; The Sobel operator is used to calculate the gradient amplitude of each pixel in the enhanced images at different scales, and the probability that each pixel in the grayscale image belongs to an edge pixel is calculated based on the difference in noise level changes of the enhanced images at each scale; The degree of irregularity of the connected domain is calculated based on the circularity and area difference of the connected domain of each frame of the grayscale image, including: ,in, Indicates the degree of irregularity of the connected domain. Represents the total number of frames of continuous grayscale images, Indicates The circularity of the connected domain on the frame grayscale image, Indicates The area of ​​the connected domain on the frame grayscale image, Represents the average area of ​​the connected domains on all continuous frame grayscale images; When the degree of irregularity of the connected domain is greater than a threshold, there is a gas leak at the location to be tested.

2. The method for detecting gas leakage of a steam boiler based on image processing according to claim 1, characterized in that: The method of using the CLAHE algorithm to obtain multiple enhanced images corresponding to each frame of grayscale image at multiple scales also includes: For each sub-region obtained at each scale of each frame of grayscale image, a corresponding adaptive contrast limiting threshold is set respectively, and the adaptive contrast limiting threshold is positively correlated with the noise level of the sub-region and negatively correlated with the sum of the noise levels of all sub-regions.

3. The method for detecting gas leakage of a steam boiler based on image processing according to claim 2, characterized in that: The adaptive contrast limiting threshold is specifically: , ,in, is the adaptive contrast limit threshold of the ith sub-region, is the initial contrast limit threshold, is the adjustment parameter of the ith sub-region, and is the noise level of the i-th and j-th sub-regions, and N is the number of sub-regions.

4. The method for detecting gas leakage of a steam boiler based on image processing according to claim 2, characterized in that: The noise level of the sub-areas includes: ,in, represents the noise level of the t-th sub-region, Indicates the number of pixels in the sub-area. represents the gray value of the w-th pixel in the sub-region, Represents the grayscale mean of all pixels in the t-th sub-region.

5. The method for detecting gas leakage of a steam boiler based on image processing according to claim 1, characterized in that: The method of obtaining a connected domain in each frame of grayscale image according to edge pixels includes: Use depth-first search or breadth-first search to get the connected domain.

6. The method for detecting gas leakage of a steam boiler based on image processing according to claim 1, characterized in that: The method of obtaining a continuous plurality of grayscale images of the position to be measured of the steam boiler comprises: A high-definition camera is used to capture continuous multiple-frame visible light images of the position to be measured in the steam boiler, and the continuous multiple-frame visible light images are converted into grayscale images.

7. The method for detecting gas leakage of a steam boiler based on image processing according to claim 1, characterized in that: When there is gas leakage at the position to be tested, an audible and visual warning is given.

8. The method for detecting gas leakage of a steam boiler based on image processing according to claim 1, characterized in that: The threshold is 0.6.

Images