Underground pipe gallery fire image adaptive enhancement method

Abstract

The invention belongs to the technical field of digital image preprocessing, and discloses an underground pipe gallery fire image adaptive enhancement method, which comprises the following steps: step 1, receiving a fire image and preprocessing the fire image to obtain a gray level histogram; step 2, calculating a segmentation threshold value of a gray level histogram; step 3, segmenting the histogram of the original image into three sub-block arrays of a low gray level region, a middle gray level region and a high gray level region by using the threshold value of the gray level histogram, and marking a pixel range corresponding to each sub-histogram; and step 4, performing enhancement processing on the three-sub-block array to obtain an enhanced fire image after enhancement reconstruction, and transmitting data of the enhanced fire image to the remote control center. The method can effectively improve the definition and contrast of the fire image in complex environments of dark light, low contrast and the like, and further achieves the quick and accurate recognition of the fire features.

Description

Technical Field

[0001] The invention belongs to the field of digital image preprocessing, and in particular relates to an underground pipe gallery fire image adaptive enhancement method. Background Art

[0002] With the acceleration of urbanization, underground pipe corridors, as an important part of urban infrastructure, bear the task of laying and maintaining a variety of key pipelines such as electricity, communications, water supply and drainage. However, due to the closed nature, small space and insufficient light of underground pipe corridors, the difficulty of monitoring and emergency response increases significantly once a fire occurs. Traditional fire monitoring methods often rely on fixed-position smoke detectors and temperature sensors, but these devices may cause false alarms or missed alarms due to factors such as dust accumulation and temperature and humidity changes in the complex and changeable underground pipe corridor environment.

[0003] In recent years, with the rapid development of image processing technology and computer vision, image-based fire detection technology has gradually attracted attention. By capturing video images in underground pipe corridors in real time with cameras and using image processing algorithms to analyze fire features in images, such as flame color, shape, and movement, early warning and accurate identification of fires can be achieved. However, the special environment of underground pipe corridors, such as dim light, low contrast, and smoke interference, poses great challenges to the clarity and contrast of fire images, which seriously affects the accurate extraction and identification of fire features.

[0004] Most existing image enhancement methods use fixed algorithm parameters or preset thresholds for image processing, which is difficult to adapt to the complex and changeable light and environmental conditions of underground pipe corridors. Therefore, a method is needed that can automatically analyze the characteristics of fire images and perform adaptive enhancement according to the unique environmental conditions of underground pipe corridors, so as to improve the clarity and contrast of fire images in complex environments, thereby realizing rapid and accurate recognition of fire characteristics. Summary of the invention

[0005] In order to solve the above technical problems, the present invention provides an adaptive enhancement method for underground pipe gallery fire images, which aims to automatically analyze the grayscale histogram of the input fire image, design and apply a set of adaptive threshold calculation rules, and perform targeted enhancement processing on the image to improve the clarity and contrast of the fire image in complex environments such as dark light and low contrast, thereby enhancing the recognition ability of fire features, while suppressing noise in non-flame areas and effectively reducing artifacts, so that the enhanced image becomes natural and clear, thereby improving the subsequent fire recognition accuracy, and providing strong support for fire monitoring and emergency response in underground pipe galleries.

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

[0007] The present invention discloses an adaptive enhancement method for underground pipe gallery fire images, which specifically comprises the following steps:

[0008] Step 1, receiving a fire image from an underground pipe gallery surveillance camera, preprocessing the fire image, and performing grayscale calculation on a grayscale histogram of the fire image;

[0009] Step 2, calculation of grayscale histogram segmentation threshold: according to the light characteristics of the underground pipe gallery (such as dark light, low contrast) and the general characteristics of the fire image, an adaptive threshold calculation rule is designed, and the threshold value of the grayscale histogram segmentation into low grayscale area, middle grayscale area and high grayscale area in step 1 is calculated by using the threshold calculation rule;

[0010] Step 3: Using the threshold of the grayscale histogram calculated in step 2, the histogram of the original image is divided into three sub-block arrays: low grayscale area, middle grayscale area and high grayscale area, and the pixel range corresponding to each sub-histogram is marked to prepare for subsequent enhancement processing;

[0011] Step 4: Perform enhancement processing on the three sub-block arrays obtained in step 3 to obtain an enhanced fire image after enhanced reconstruction, and transmit the data of the enhanced fire image to the remote control center.

[0012] A further improvement of the present invention is that step 1 specifically comprises the following steps:

[0013] Step 1.1, calculate the gray level histogram h(i) of the gray level input fire image I, that is, the frequency value array of gray level i, where h is the array name, i is the gray level value, 1≤i≤256;

[0014] Step 1.2, put the values ​​of the gray level histogram h(i)>0 into the frequency value array H(n) that is not 0 in the order of gray level value i from small to large:

[0015] H(n)=h(i)

[0016] Where h(i)>0, H is the array name, n is the position value of the frequency value array H(n), 1≤n≤256;

[0017] Step 1.3, for the value of h(i)>0, put i into id(n) in the order of grayscale value i from small to large, the expression is: id(n)=i, h(i)>0, where id is the array name, id(n) is the array of grayscale value i, n is the position value of the array id(n) of grayscale value i, 1≤n≤256.

[0018] A further improvement of the present invention is that in step 2, the following steps are specifically included:

[0019] Step 2.1, calculate the maximum value Hmax, minimum value Hmin and standard deviation Hd in the frequency value array H(n):

[0020] Hmax=max(H(n));

[0021] Hmin=min(H(n));

[0022] Hd = std(H(n));

[0023] According to Hmax, Hmin, and Hd, the upper frequency limit HL and lower frequency limit HU of the gray level histogram h(i) of the input fire image I are calculated:

[0024] HL=Hmin+Hd

[0025] HU=Hmax-Hd

[0026] At the same time, find the position value lookup table idmax corresponding to the maximum value Hmax in the array gray level histogram h(i):

[0027] idmax=find(h(i)==Hmax

[0028] Among them, max() and min() are the maximum and minimum value finding functions of the MATLAB language respectively, and find() is a numerical search function;

[0029] Step 2.2, based on HL and HU, calculate the intermediate arrays h1(i) and h2(i):

[0030] h1(i)=│h(i)-HL│

[0031] h2(i)=│h(i)-HU│

[0032] And calculate the four minimum values ​​of h1(i) and h2(i) in the two grayscale value intervals [1idmax] and [idmax+1256]:

[0033] h1_min1=min(h1(1:idmax))

[0034] h1_min2=min(h1(idmax+1:256))

[0035] h2_min1=min(h2(1:idmax))

[0036] h2_min2=min(h2(idmax+1:256));

[0037] Step 2.3, in the two intervals [1idmax] and [idmax+1256] of the grayscale values ​​of the arrays h1(i) and h2(i), find the corresponding grayscale threshold values ​​A1, A2, B1, B2 of the four minimum values ​​h1_min1, h1_min2, h2_min1, and h2_min2:

[0038] A1==find(h1(1:idmax)==h1_min1)

[0039] A2=find(h1(idmax+1:256)==h1_min2)

[0040] B1==find(h2(1:idmax)==h2_min1)

[0041] B2=find(h2(idmax+1:256)==h2_min2).

[0042] A further improvement of the present invention is that the step 3 specifically comprises the following steps:

[0043] Step 3.1, construct a new array C = [A1 A2 B1 B2], calculate the upper segmentation gray level threshold XL and the lower segmentation gray level threshold XU:

[0044] XL=round((A1+B1) / 2)

[0045] XU=round((A2+B2) / 2)

[0046] Find the upper segmentation gray level threshold XL and the lower segmentation gray level threshold XU in the array id(n) of the gray level value i:

[0047] id_XL=find(id(n)==XL

[0048] id_XU=find(id(n)==XU);

[0049] Step 3.2: Split the arrays H(n) and id(n) into three corresponding sub-block arrays (H1(j), idH1(j); H2(k), idH2(k); H3(q), idH3(q)) according to the position segmentation thresholds id_XL and id_XU. The segmentation mathematical expression is:

[0050]

[0051]

[0052]

[0053] Among them, end represents the last digit of the array position value, and j, k, and q are variables.

[0054] A further improvement of the present invention is that in step 4, the sub-histogram obtained in step 3 is enhanced according to the region, specifically:

[0055] Low grayscale area: Use mapping method to improve dark details and enhance the visibility of potential fire features;

[0056] Middle gray area: Maintain or fine-tune to maintain the natural transition and details of the image;

[0057] High grayscale area: appropriately compress the bright area to avoid overexposure, while enhancing the contrast of bright features such as flames.

[0058] A further improvement of the present invention is that in step 4, the acquisition of the enhanced fire image is specifically: re-mapping the pixel values ​​of the enhanced low grayscale area, the middle grayscale area and the high grayscale area back to the corresponding positions of the original image, assigning a new grayscale value to each pixel in the image by interpolation or direct assignment, and generating a global enhanced fire image. In the process of fire image enhancement or reconstruction, it is necessary to ensure that the transition between different sub-areas is natural to avoid obvious splicing marks or artifacts.

[0059] A further improvement of the present invention is that the step 4 specifically comprises the following steps:

[0060] Step 4.1, calculate the three sub-block arrays (H1(j), idH1(j); H2(k), idH2(k); H3(q), idH3(q)) and obtain the new grayscale value array (idH1_x(j), idH2_x(k), idH3_x(q)) after mathematical transformation, which is the grayscale value of the enhanced output image. The transformation expression is:

[0061]

[0062] Among them, idH1(1), idH2(1), and idH3(1) are the first values ​​of the corresponding arrays respectively, idH1(end), idH2(end), and idH3(end) are the last values ​​of the corresponding arrays, i.e., the tail values, respectively, length(idH1), length(idH2), and length(idH3) are the lengths of the corresponding arrays respectively, and the variable j, k, and q vary in the ranges of 1≤j≤length(idH1)), 1≤k≤length(idH2), and 1≤q≤length(idH3), respectively.

[0063] The beneficial effects of the present invention are:

[0064] The present invention can significantly improve the brightness of the image through adaptive threshold calculation and enhancement processing, so that the fire characteristics can be clearly seen even in a dark light environment, solving the technical problem in the prior art that the images captured by the camera are often dim due to the closedness and space limitations of underground pipe galleries, which makes it difficult for natural light to enter.

[0065] The present invention can enhance the contrast of the image through histogram segmentation and targeted enhancement processing, making the fire characteristics more distinct, facilitating subsequent analysis and identification, and solving the problem of blurred image details caused by low contrast in the prior art.

[0066] The present invention automatically analyzes the grayscale histogram of the fire image and designs an adaptive threshold calculation rule according to the unique environmental conditions of the underground pipe gallery, thereby achieving effective enhancement of the fire image under different environmental conditions, improving the adaptability and robustness of the method, and solving the technical defect that the traditional image enhancement method often adopts fixed algorithm parameters or preset thresholds and is difficult to adapt to the complex and changeable environmental conditions of the underground pipe gallery. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 It is a flow chart of the present invention.

[0068] Figure 2 It is a processing flow chart of the method of the present invention.

[0069] Figure 3 It is a comparison diagram of the fire enhancement effect of the underground pipe gallery of the present invention.

[0070] Figure 4 It is the effect diagram of the histogram of the present invention. DETAILED DESCRIPTION

[0071] The following will disclose the embodiments of the present invention with drawings. For the purpose of clear description, many practical details will be described together in the following description. However, it should be understood that these practical details should not be used to limit the present invention. That is to say, in some embodiments of the present invention, these practical details are not necessary.

[0072] The present invention is based on the digital grayscale image histogram, and according to the frequency standard deviation of the input image grayscale histogram, two segmentation thresholds of the histogram are adaptively calculated, and the input image histogram is divided into three sub-histograms, and the grayscale adaptive enhancement is performed on the first two sub-histograms, and the third sub-histogram is balanced and grayscale adaptive mixed enhancement is performed to calculate the enhanced image;

[0073] On the one hand, the present invention can effectively improve the image contrast while greatly reducing the whitening phenomenon of the picture caused by over-enhancement, and can also effectively reduce artifacts, so that the enhanced image becomes natural and clear; on the other hand, it can also suppress noise in non-flame areas and improve the subsequent fire recognition accuracy.

[0074] like Figure 1-2 As shown, the present invention is an adaptive enhancement method for underground pipe gallery fire images, and the enhancement method specifically comprises the following steps:

[0075] Step 1: Receive the fire image from the underground pipe gallery surveillance camera and preprocess the fire image. The preprocessing includes denoising, resizing, etc., in preparation for grayscale histogram analysis, calculate the grayscale histogram of the fire image, and provide basic data for subsequent processing.

[0076] In this step 1, the MATLAB programming language is used for description, and the specific steps include:

[0077] Step 1.1, calculate the gray level histogram h(i) of the gray level input fire image I, that is, the frequency value array of gray level i, where h is the array name, i is the gray level value, 1≤i≤256;

[0078] Step 1.2, put the values ​​of the gray level histogram h(i)>0 into the frequency value array H(n) that is not 0 in the order of gray level value i from small to large:

[0079] H(n)=h(i)

[0080] Where h(i)>0, H is the array name, n is the position value of the frequency value array H(n), 1≤n≤256;

[0081] Step 1.3, for the value of h(i)>0, put i into id(n) in the order of grayscale value i from small to large, the expression is: id(n)=i, h(i)>0, where id is the array name, id(n) is the array of grayscale value i, n is the position value of the array id(n) of grayscale value i, 1≤n≤256.

[0082] Step 2, calculation of grayscale histogram segmentation threshold: according to the light characteristics of the underground pipe gallery, such as dark light, low contrast and general characteristics of fire images, an adaptive threshold calculation rule is designed, and the threshold value of the grayscale histogram segmentation into low grayscale area, middle grayscale area and high grayscale area in step 1 is calculated by using the threshold calculation rule;

[0083] The histogram segmentation threshold calculation unit specifically includes the following steps:

[0084] Step 2.1, calculate the maximum value Hmax, minimum value Hmin and standard deviation Hd in the frequency value array H(n):

[0085] Hmax=max(H(n));

[0086] Hmin=min(H(n));

[0087] Hd = std(H(n));

[0088] According to Hmax, Hmin, and Hd, the upper frequency limit HL and lower frequency limit HU of the gray level histogram h(i) of the input fire image I are calculated:

[0089] HL=Hmin+Hd

[0090] HU=Hmax-Hd

[0091] At the same time, find the position value lookup table idmax corresponding to the maximum value Hmax in the array gray level histogram h(i):

[0092] idmax=find(h(i)==Hmax

[0093] Among them, max() and min() are the maximum and minimum value finding functions of the MATLAB language respectively, and find() is a numerical search function;

[0094] Step 2.2, based on HL and HU, calculate the intermediate arrays h1(i) and h2(i):

[0095] h1(i)=│h(i)-HL│

[0096] h2(i)=│h(i)-HU│

[0097] And calculate the four minimum values ​​of h1(i) and h2(i) in the two grayscale value intervals [1idmax] and [idmax+1256]:

[0098] h1_min1=min(h1(1:idmax))

[0099] h1_min2=min(h1(idmax+1:256))

[0100] h2_min1=min(h2(1:idmax))

[0101] h2_min2=min(h2(idmax+1:256));

[0102] Step 2.3, in the two intervals [1idmax] and [idmax+1256] of the grayscale values ​​of the arrays h1(i) and h2(i), find the corresponding grayscale values ​​A1, A2, B1, B2 of the four minimum values ​​h1_min1, h1_min2, h2_min1, h2_min2:

[0103] A1==find(h1(1:idmax)==h1_min1)

[0104] A2=find(h1(idmax+1:256)==h1_min2)

[0105] B1==find(h2(1:idmax)==h2_min1)

[0106] B2=find(h2(idmax+1:256)==h2_min2).

[0107] Step 3: Using the threshold of the grayscale histogram calculated in step 2, the histogram of the original image is divided into three sub-block arrays: low grayscale area, middle grayscale area and high grayscale area, and the pixel range corresponding to each sub-histogram is marked to prepare for subsequent enhancement processing, which specifically includes the following steps:

[0108] Step 3.1, construct a new array C = [A1 A2 B1 B2], calculate the upper segmentation gray level threshold XL and the lower segmentation gray level threshold XU:

[0109] XL=round((A1+B1) / 2)

[0110] XU=round((A2+B2) / 2)

[0111] Find the upper segmentation gray level threshold XL and the lower segmentation gray level threshold XU in the array id(n) of the gray level value i:

[0112] id_XL=find(id(n)==XL

[0113] id_XU=find(id(n)==XU);

[0114] Step 3.2: Split the arrays H(n) and id(n) into three corresponding sub-block arrays (H1(j), idH1(j); H2(k), idH2(k); H3(q), idH3(q)) according to the position segmentation thresholds id_XL and id_XU. The segmentation mathematical expression is:

[0115]

[0116]

[0117]

[0118] Among them, end represents the last digit of the array position value, and j, k, and q are variables.

[0119] Step 4: Enhance the three sub-block arrays obtained in step 3 to obtain an enhanced fire image after enhanced reconstruction, and transmit the data of the enhanced fire image to the remote control center. Perform targeted enhancement processing on the obtained sub-histogram according to the region, specifically:

[0120] Low grayscale area: Use mapping method to improve dark details and enhance the visibility of potential fire features;

[0121] Middle gray area: Maintain or fine-tune to maintain the natural transition and details of the image;

[0122] High grayscale area: appropriately compress the bright area to avoid overexposure, while enhancing the contrast of bright features such as flames.

[0123] The pixel values ​​of the enhanced low-grayscale area, middle-grayscale area, and high-grayscale area are remapped back to the corresponding positions of the original image, and a new grayscale value is assigned to each pixel in the image by interpolation or direct assignment to generate a global enhanced fire image. In the process of fire image enhancement or reconstruction, it is necessary to ensure that the transition between different sub-areas is natural to avoid obvious splicing marks or artifacts.

[0124] The reconstructed enhanced image is displayed in real time on the monitoring screen so that on-site personnel can immediately observe the changes at the fire scene. At the same time, the enhanced image data is transmitted to the remote control center via wired or wireless means for further analysis and decision-making by professionals. During the transmission process, data compression and encryption may need to be considered to ensure the efficiency and security of data transmission.

[0125] The specific steps include:

[0126] Step 4.1, calculate the three sub-block arrays (H1(j), idH1(j); H2(k), idH2(k); H3(q), idH3(q)) and obtain the new grayscale value array (idH1_x(j), idH2_x(k), idH3_x(q)) after mathematical transformation, which is the grayscale value of the enhanced output image. The transformation expression is:

[0127]

[0128] Among them, idH1(1), idH2(1), and idH3(1) are the first values ​​of the corresponding arrays respectively, idH1(end), idH2(end), and idH3(end) are the last values ​​of the corresponding arrays, i.e., the tail values, respectively, length(idH1), length(idH2), and length(idH3) are the lengths of the corresponding arrays respectively, and the variable j, k, and q vary in the ranges of 1≤j≤length(idH1)), 1≤k≤length(idH2), and 1≤q≤length(idH3), respectively.

[0129] The method of the present invention is mainly based on the histogram of a digital grayscale image. The upper and lower segmentation thresholds are calculated according to the threshold calculation rules of the input image histogram. The histogram of the input image is automatically segmented into three sub-histograms by the upper and lower segmentation thresholds, and the enhanced images are calculated and output for the three sub-histograms respectively.

[0130] In order to verify the present invention, the histograms and enhancement effects of the classical histogram enhancement method and the processing method of the present invention are compared. The enhancement effect comparison diagram is shown in FIG. Figure 3 As shown, Figure 3 (a) is the original image, Figure 3 (b) is the image of the classic histogram enhancement method. Figure 3 (c) is the image enhanced by the method of the present application.

[0131] Figure 4 (a) is the original image histogram, Figure 4 (b) is the histogram enhanced by the classic histogram. Figure 4 (c) Histogram enhanced for this application.

[0132] According to the attached Figure 3 and Figure 4 As shown in Figure 3(a), after the original image is enhanced by the classic histogram and the method of the present invention, the image processed by the classic histogram is brighter overall and has a whitening phenomenon, which can easily lead to misjudgment in subsequent fire identification. Figure 4 The classic histogram-enhanced image in (b) also shows that the histogram interval is too large, resulting in excessive contrast and serious loss of detail information.

[0133] After the enhancement of the present invention Figure 3 (c) The brightness of the picture does not change much. Only the flame area is highlighted and enhanced, while the rest is weakened and enhanced. The overall picture and poem appear clearer and more natural.

[0134] The above description is only an embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.

Claims

1. An adaptive enhancement method for underground pipe gallery fire images, characterized in that: The enhancement method specifically comprises the following steps: Step 1, receiving a fire image from an underground pipe gallery surveillance camera, preprocessing the fire image, and performing grayscale calculation on a grayscale histogram of the fire image; Step 2, calculate the segmentation threshold of the gray level histogram: design an adaptive threshold calculation rule, and use the threshold calculation rule to calculate the threshold for segmenting the gray level histogram in step 1 into low gray area, middle gray area and high gray area; Step 3: Using the threshold of the grayscale histogram calculated in step 2, the histogram of the original image is divided into three sub-block arrays: low grayscale area, middle grayscale area and high grayscale area, and the pixel range corresponding to each sub-histogram is marked to prepare for subsequent enhancement processing; Step 4: Perform enhancement processing on the three sub-block arrays obtained in step 3 to obtain an enhanced fire image after enhanced reconstruction, and transmit the data of the enhanced fire image to the remote control center.

2. The method for adaptively enhancing underground pipe gallery fire images according to claim 1 is characterized in that: The step 1 specifically includes the following steps: Step 1.1, calculate the gray level histogram h(i) of the gray level input fire image I, that is, the frequency value array of gray level i, where h is the array name, i is the gray level value, 1≤i≤256; Step 1.2, put the values ​​of the gray level histogram h(i)>0 into the frequency value array H(n) that is not 0 in the order of gray level value i from small to large: H(n)=h(i) Where h(i)>0, H is the array name, n is the position value of the frequency value array H(n), 1≤n≤256; Step 1.3, for the value of h(i)>0, put i into id(n) in the order of grayscale value i from small to large, the expression is: id(n)=i, h(i)>0, where id is the array name, id(n) is the array of grayscale value i, n is the position value of the array id(n) of grayscale value i, 1≤n≤256.

3. The method for adaptively enhancing underground pipe gallery fire images according to claim 2 is characterized in that: In step 2, the following steps are specifically included: Step 2.1, calculate the maximum value Hmax, minimum value Hmin and standard deviation Hd in the frequency value array H(n): Hmax=max(H(n)); Hmin=min(H(n)); Hd = std(H(n)); According to Hmax, Hmin, and Hd, the upper frequency limit HL and lower frequency limit HU of the gray level histogram h(i) of the input fire image I are calculated: HL=Hmin+Hd HU=Hmax-Hd At the same time, find the position value lookup table idmax corresponding to the maximum value Hmax in the array gray level histogram h(i): idmax=find(h(i)==Hmax Among them, max() and min() are the maximum and minimum value finding functions respectively, and find() is the value finding function; Step 2.2, based on HL and HU, calculate the intermediate arrays h1(i) and h2(i): h1(i)=│h(i)-HL│ h2(i)=│h(i)-HU│ And calculate the four minimum values ​​of h1(i) and h2(i) in the two grayscale value intervals [1idmax] and [idmax+1256]: h1_min1=min(h1(1:idmax)) h1_min2=min(h1(idmax+1:256)) h2_min1=min(h2(1:idmax)) h2_min2=min(h2(idmax+1:256)); Step 2.3, in the two intervals [1idmax] and [idmax+1256] of the grayscale values ​​of the arrays h1(i) and h2(i), find the corresponding grayscale values ​​A1, A2, B1, B2 of the four minimum values ​​h1_min1, h1_min2, h2_min1, h2_min2: A1==find(h1(1:idmax)==h1_min1) A2=find(h1(idmax+1:256)==h1_min2) B1==find(h2(1:idmax)==h2_min1) B2=find(h2(idmax+1:256)==h2_min2).

4. The method for adaptively enhancing underground pipe gallery fire images according to claim 3 is characterized in that: Described step 3 specifically comprises the following steps: Step 3.1, construct a new array C = [A1 A2 B1 B2], calculate the upper segmentation gray level threshold XL and the lower segmentation gray level threshold XU: XL=round((A1+B1) / 2) XU=round((A2+B2) / 2) Find the upper segmentation gray level threshold XL and the lower segmentation gray level threshold XU in the array id(n) of the gray level value i: id_XL=find(id(n)==XL id_XU=find(id(n)==XU); Step 3.2: Split the arrays H(n) and id(n) into three corresponding sub-block arrays (H1(j), idH1(j); H2(k), idH2(k); H3(q), idH3(q)) according to the position segmentation thresholds id_XL and id_XU. The segmentation mathematical expression is: Among them, end represents the last digit of the array position value, and j, k, and q are variables.

5. The method for adaptively enhancing underground pipe gallery fire images according to claim 4 is characterized in that: In step 4, the sub-histogram obtained in step 3 is enhanced according to the region, specifically: Low grayscale area: Use mapping method to improve dark details and enhance the visibility of potential fire features; Middle gray area: Maintain or adjust to keep the natural transition and details of the image; High Gray Area: Compresses bright areas while enhancing the contrast of the bright features of the flame.

6. The method for adaptively enhancing underground pipe gallery fire images according to claim 5 is characterized in that: In step 4, the enhanced fire image is obtained by remapping the pixel values ​​of the enhanced low-grayscale area, middle-grayscale area and high-grayscale area back to the corresponding positions of the original image, assigning a new grayscale value to each pixel in the image, and generating a global enhanced fire image.

7. The method for adaptively enhancing underground pipe gallery fire images according to claim 6 is characterized in that: The step 4 specifically comprises the following steps: Step 4.1, calculate the three sub-block arrays (H1(j), idH1(j); H2(k), idH2(k); H3(q), idH3(q)) and obtain the new grayscale value array (idH1_x(j), idH2_x(k), idH3_x(q)) after mathematical transformation, which is the grayscale value of the enhanced output image. The transformation expression is: Among them, idH1(1), idH2(1), and idH3(1) are the first values ​​of the corresponding arrays respectively, idH1(end), idH2(end), and idH3(end) are the last values ​​of the corresponding arrays, i.e., the tail values, respectively, length(idH1), length(idH2), and length(idH3) are the lengths of the corresponding arrays respectively, and the variable j, k, and q vary in the ranges of 1≤j≤length(idH1)), 1≤k≤length(idH2), and 1≤q≤length(idH3), respectively.

Images