Fire safety early warning method for integrated pipe gallery based on image processing

Abstract

The invention relates to the field of fire warning technology for an integrated pipe gallery, and comprises the following steps: obtaining pipe gallery information of a current integrated pipe gallery fire management area, and setting fire video monitoring points; constructing a visual map for monitoring the entire area of ​​the integrated pipe gallery, screening out video frame images containing suspicious areas through video data of each fire video monitoring point, judging the number of flame corners through three steps of setting a recognition interval of flame corners, an average value of corners of different segments within the recognition interval, and an approximate triangular shape of flame corners, thereby improving the accuracy of judging flame corners; judging whether a fire occurs in an area in the video frame image based on the number of flame corners in the suspicious area, and judging the type of fire; and constructing a regression model to obtain a fire change trend prediction result, and visually displaying the fire change trend prediction result through a visual map for monitoring the entire area of ​​the integrated pipe gallery, thereby making the way of displaying fire data clearer and more intuitive.

Description

Technical Field

[0001] The present invention relates to the technical field of fire warning, and in particular to an image processing-based comprehensive pipe gallery fire safety warning method. Background Art

[0002] The integrated pipe gallery, also known as the common trench, refers to a public tunnel underground in the city for laying municipal pipelines such as electricity, communications, radio and television, water supply, drainage, heat and gas. It is a modern, scientific and intensive urban infrastructure. The design of the integrated pipe gallery provides convenience for optimizing urban space, improving urban appearance, installing, maintaining and removing various municipal pipelines. The fire safety warning of the integrated pipe gallery also plays an increasingly important role.

[0003] Traditional video image fire detection and recognition technology usually uses the shape of sharp corners in a single frame image to judge the number of flame sharp corners, and finally judges whether there is fire information in a single frame image based on the number of flame sharp corners. This method has poor robustness and is easily affected by interference such as sunlight, lighting, and color patterns in practical applications. It does not take into account the connection and coupling relationship between multiple frames of images, and has low accuracy in judging the occurrence of fire and high false alarm rate. In order to solve the above technical problems, a fire safety early warning method for a comprehensive pipe gallery based on image processing is provided. Summary of the invention

[0004] In order to solve the above technical problems, the purpose of the present invention is to provide a fire safety early warning method for an integrated pipe gallery based on image processing, comprising the following steps:

[0005] Step S1: Obtain the pipe gallery information of the current integrated pipe gallery fire management area, and set fire video monitoring points according to the pipe gallery information;

[0006] Step S2: construct a visual map of the entire area of ​​the integrated pipe gallery, filter out video frame images containing suspicious areas through the video data of each fire video monitoring point, and determine the number of flame corners in the suspicious area by setting the identification interval of flame corners, the average value of the corners at different stages in the identification interval, and the approximate triangular shape of the flame corners;

[0007] Step S3: judging whether a fire occurs in the area in the video frame image based on the number of flame corners in the suspicious area, and judging the type of fire;

[0008] Step S4: construct a regression model through the video frame images of the fire, obtain the fire change trend prediction results, and visualize the fire change trend prediction results through the comprehensive pipe gallery full-area monitoring visual diagram.

[0009] Furthermore, in order to obtain the pipe gallery information of the current integrated pipe gallery fire management area, the process of setting the fire video monitoring point according to the pipe gallery information includes:

[0010] Obtain the municipal pipeline laying characteristics of the current integrated pipe gallery fire management area, extract the pipe gallery information according to the municipal pipeline laying characteristics, split the integrated pipe gallery fire management area according to the pipe gallery information, and divide it into several integrated pipe gallery sub-areas;

[0011] According to the pipeline corridor information and location characteristics of each comprehensive pipeline corridor sub-area, the evaluation index is selected, the index weight matrix of the evaluation index is set, and the membership matrix of each comprehensive pipeline corridor sub-area for the preset importance level is obtained through the fuzzy comprehensive evaluation method;

[0012] The importance level of each integrated pipe gallery sub-area is obtained according to the membership matrix and the indicator weight matrix, the number of fire video monitoring points is determined according to the importance level of the integrated pipe gallery sub-area, and the distribution of fire video monitoring points in the integrated pipe gallery sub-area is determined according to the effective coverage area of ​​the fire video monitoring points and the location characteristics of the integrated pipe gallery sub-area.

[0013] Furthermore, the process of constructing a visual map of the entire area of ​​the integrated pipe gallery includes:

[0014] The physical entities of the municipal pipelines in the current fire management area of ​​the integrated pipe gallery are obtained, and a map of the integrated pipe gallery is constructed according to the location characteristics and pipe gallery connection relationship between the sub-areas of the integrated pipe gallery in the current fire management area of ​​the integrated pipe gallery. The physical entities of the municipal pipelines are subjected to three-dimensional modeling and mapped to the integrated pipe gallery map to obtain a three-dimensional model;

[0015] The multi-source video data of each fire video monitoring point in each integrated pipe gallery sub-area is obtained, and the multi-source video data is preprocessed in data format to obtain twin data. According to the assembly connection relationship between each fire video monitoring point and each integrated pipe gallery sub-area in the physical space, the twin data is matched with the three-dimensional model to obtain a visual monitoring diagram of the entire integrated pipe gallery area.

[0016] Furthermore, the process of screening out video frame images containing suspicious areas from the video data of each fire video monitoring point includes:

[0017] Obtain video data of each fire video monitoring point, convert the obtained video data into continuous video frame images, set a reference video frame image of each fire video monitoring point, subtract the pixel values ​​of the video frame image and the reference video frame image at corresponding positions to obtain a difference value, convert the difference value into a pixel value of a binary image, and generate a binary image;

[0018] Set two pixel traversal pointers and start traversal from the first pixel and the last pixel in the binary image area at the same time;

[0019] Mark the pixel point area in the binary image whose pixel value is greater than the preset pixel threshold as a suspicious area;

[0020] The pixel point area in the binary image whose pixel value is less than or equal to the preset pixel threshold is marked as a normal area.

[0021] Furthermore, the process of determining the number of flame corners in the suspicious area by setting the identification interval of the flame corners, the average value of the corners at different stages in the identification interval, and the approximate triangular shape of the flame corners includes:

[0022] Select the pixel point at the center of the suspicious area in the binary image as the central pixel point, establish a two-dimensional coordinate system with the central pixel point as the origin, map all the pixels in the suspicious area to the two-dimensional coordinate system, and obtain the Euclidean distance between each boundary pixel point on the edge of the suspicious area and the central pixel point;

[0023] Construct a one-way linked list, randomly select a boundary pixel point on the edge of the suspicious area in the binary image, and enter the two-dimensional coordinates and Euclidean distances of all boundary pixels on the edge of the suspicious area clockwise into the one-way linked list in sequence, starting from the boundary pixel point, and determine the identification interval of the flame tip as n boundary pixels according to the total number of boundary pixels, and mark the boundary pixels in the one-way linked list whose Euclidean distance is greater than the Euclidean distance of n / 2 boundary pixels in the left and right neighborhoods as possible flame tips;

[0024] The sum of the Euclidean distances S1, S2 and S3 of the n / 8, n / 4 and n / 2 boundary pixel points in the left and right areas of the possible flame corner are obtained respectively. When S1<S2<S3, the possible flame corner is converted into a suspected flame corner.

[0025] An approximate triangle is constructed with the two-dimensional coordinates of the suspected flame corner as the vertex and the two-dimensional coordinates of n boundary pixels in the flame corner identification interval as the base, and the tangent value of the standard vertex angle half-angle of the approximate triangle of the flame corner is set. When the tangent value of the vertex angle half-angle of the approximate triangle of the suspected flame corner is less than or equal to the tangent value of the standard vertex angle half-angle, the suspected flame corner is converted into a flame corner.

[0026] Furthermore, the process of determining whether a fire occurs in the area in the video frame image based on the number of flame corners in the suspicious area and determining the type of the fire includes:

[0027] Select the number of flame corners of the video frame image at the tth second, and compare the number of flame corners of the video frame image with a preset number of flame corners;

[0028] When the number of flame sharp corners is greater than a preset number of flame sharp corners, the video frame image is marked as a fire source image, and a suspicious area in the video frame image is marked as a flame area;

[0029] When the number of flame sharp corners is less than or equal to the preset number of flame sharp corners, marking the video frame image as an image of other interference sources;

[0030] If the video frame image at the t second is a fire source image, the number of flame corners in the video frame image is averaged by difference with the number of flame corners in the video frame image at the t-1 second, the number of flame corners in the video frame image at the t-2 second, and the number of flame corners in the video frame image at the t-3 second, respectively, to obtain the average value of the change in the corners of the video frame image at the t second and three consecutive adjacent video frame images;

[0031] Set the average threshold of sharp angle change, and compare the average value of the sharp angle change of the video frame image at the tth second with the average threshold of the sharp angle change;

[0032] If the average value of the sharp angle change is greater than the average threshold value of the sharp angle change, the fire source in the video frame image is marked as an out-of-control fire source;

[0033] If the average value of the sharp angle change is less than or equal to the average threshold of the sharp angle change, the fire source in the video frame image is marked as a stable fire source.

[0034] Furthermore, the process of building a regression model through the video frame images of the fire and obtaining the fire change trend prediction results includes:

[0035] The continuous video frame images of the fire source being the uncontrolled fire source in the fire source image are sampled at equal intervals to obtain k sample video frame images;

[0036] Generate a flame area according to the number of pixels contained in the flame area in the video frame image, and generate a centroid of the flame area according to the pixel values ​​of the pixels contained in the flame area in the video frame image;

[0037] The flame area and the flame area centroid of k sample video frame images are curve-fitted using a regression model to obtain a flame area growth curve and a flame area centroid movement curve respectively;

[0038] The flame area change trend and the flame spreading direction trend in the next video frame image are obtained according to the flame area growth curve and the flame area centroid movement curve.

[0039] Furthermore, the process of visually displaying the fire change trend prediction results through the comprehensive pipe gallery full area monitoring visual map includes:

[0040] Obtain the analysis result of the video frame image of each fire video monitoring point. If the analysis result of the video frame image is a fire source image, and the fire source is an out-of-control fire source, the comprehensive pipe gallery full-area monitoring visual map will display the comprehensive pipe gallery sub-area where the out-of-control fire source is located as a red alarm area;

[0041] If the analysis result of the video frame image is a fire source image, and the fire source is a stable fire source, the comprehensive pipe gallery full-area monitoring visual map will display the comprehensive pipe gallery sub-area where the out-of-control fire source is located as an orange alarm area;

[0042] The visual graph of the comprehensive pipe gallery full-area monitoring obtains the fire spread arrival time and fire spread range of the non-alarm area according to the flame area change trend and flame spread direction trend of the red alarm area, and sets the non-alarm area within the fire spread range as the time-divided early warning area according to the fire spread arrival time.

[0043] Compared with the prior art, the present invention has the following beneficial effects: compared with the traditional video image fire detection and recognition technology which usually uses the shape of the sharp angle in a single frame image to judge the number of flame sharp angles, the present invention sets the identification interval of the flame sharp angle, the average value of the sharp angle in different segments within the identification interval and the approximate triangular shape of the flame sharp angle to judge the number of flame sharp angles, which greatly reduces the interference of sunlight, light, color patterns, etc. and improves the accuracy of judging the number of flame sharp angles;

[0044] On the other hand, by comparing the number of flame corners between multiple frame images, the flame type is obtained, and the fire change trend prediction results are visualized through the comprehensive pipeline corridor full-area monitoring visual graph, achieving the effect of multiple uses of one graph. Compared with traditional fire data statistics tables, the comprehensive pipeline corridor full-area monitoring visual graph of the present invention displays fire data in a clearer and more intuitive way. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a schematic diagram of the integrated pipe gallery fire safety warning method based on image processing according to an embodiment of the present application. Implementation

[0046] The following is a clear and complete description of the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0047] like Figure 1 As shown, the fire safety early warning method of the integrated pipe gallery based on image processing includes the following steps:

[0048] Step S1: Obtain the pipe gallery information of the current integrated pipe gallery fire management area, and set fire video monitoring points according to the pipe gallery information;

[0049] Step S2: construct a visual map of the entire area of ​​the integrated pipe gallery, filter out video frame images containing suspicious areas through the video data of each fire video monitoring point, and determine the number of flame corners in the suspicious area by setting the identification interval of flame corners, the average value of the corners at different stages in the identification interval, and the approximate triangular shape of the flame corners;

[0050] Step S3: judging whether a fire occurs in the area in the video frame image based on the number of flame corners in the suspicious area, and judging the type of fire;

[0051] Step S4: construct a regression model through the video frame images of the fire, obtain the fire change trend prediction results, and visualize the fire change trend prediction results through the comprehensive pipe gallery full-area monitoring visual diagram.

[0052] It should be further explained that, in the specific implementation process, the process of obtaining the pipe gallery information of the current integrated pipe gallery fire management area and setting the fire video monitoring points according to the pipe gallery information includes:

[0053] Obtain the municipal pipeline laying characteristics of the current integrated pipe gallery fire management area, extract the pipe gallery information according to the municipal pipeline laying characteristics, split the integrated pipe gallery fire management area according to the pipe gallery information, and divide it into several integrated pipe gallery sub-areas; the integrated pipe gallery sub-areas include trunk integrated pipe gallery, branch integrated pipe gallery, sub-branch integrated pipe gallery and cable integrated pipe gallery; the pipe gallery information includes the number of municipal pipelines of various types, the location characteristics of the integrated pipe gallery and the size of the internal space of the integrated pipe gallery;

[0054] According to the pipeline corridor information and location characteristics of each comprehensive pipeline corridor sub-area, the evaluation index is selected, the index weight matrix of the evaluation index is set, and the membership matrix of each comprehensive pipeline corridor sub-area for the preset importance level is obtained through the fuzzy comprehensive evaluation method;

[0055] The importance level of each integrated pipe gallery sub-area is obtained according to the membership matrix and the indicator weight matrix, the number of fire video monitoring points is determined according to the importance level of the integrated pipe gallery sub-area, and the distribution of fire video monitoring points in the integrated pipe gallery sub-area is determined according to the effective coverage area of ​​the fire video monitoring points and the location characteristics of the integrated pipe gallery sub-area.

[0056] It should be further explained that, in the specific implementation process, the process of building a visual monitoring map of the entire area of ​​the integrated pipeline corridor includes:

[0057] The physical entities of the municipal pipelines in the current fire management area of ​​the integrated pipe gallery are obtained, and a map of the integrated pipe gallery is constructed according to the location characteristics and pipe gallery connection relationship between the sub-areas of the integrated pipe gallery in the current fire management area of ​​the integrated pipe gallery. The physical entities of the municipal pipelines are subjected to three-dimensional modeling and mapped to the integrated pipe gallery map to obtain a three-dimensional model;

[0058] The multi-source video data of each fire video monitoring point in each integrated pipe gallery sub-area is obtained, and the multi-source video data is preprocessed in data format to obtain twin data. According to the assembly connection relationship between each fire video monitoring point and each integrated pipe gallery sub-area in the physical space, the twin data is matched with the three-dimensional model to obtain a visual monitoring diagram of the entire integrated pipe gallery area.

[0059] It should be further explained that, in the specific implementation process, the process of screening out video frame images containing suspicious areas through the video data of each fire video monitoring point includes:

[0060] Obtain video data of each fire video monitoring point, convert the obtained video data into continuous video frame images, set a reference video frame image of each fire video monitoring point, subtract the pixel values ​​of the video frame image and the reference video frame image at corresponding positions to obtain a difference value, convert the difference value into a pixel value of a binary image, and generate a binary image;

[0061] Set two pixel traversal pointers and start traversal from the first pixel and the last pixel in the binary image area at the same time;

[0062] Mark the pixel point area in the binary image whose pixel value is greater than the preset pixel threshold as a suspicious area;

[0063] The pixel point area in the binary image whose pixel value is less than or equal to the preset pixel threshold is marked as a normal area.

[0064] It should be further explained that, in the specific implementation process, the process of determining the number of flame corners in the suspicious area by setting the identification interval of the flame corners, the average value of the corners at different stages in the identification interval, and the approximate triangular shape of the flame corners includes:

[0065] Select the pixel point at the center of the suspicious area in the binary image as the central pixel point, establish a two-dimensional coordinate system with the central pixel point as the origin, map all the pixels in the suspicious area to the two-dimensional coordinate system, and obtain the Euclidean distance between each boundary pixel point on the edge of the suspicious area and the central pixel point;

[0066] A one-way linked list is constructed, a boundary pixel point at the edge of the suspicious area in the binary image is randomly selected, and the two-dimensional coordinates and Euclidean distances of all boundary pixel points at the edge of the suspicious area are sequentially entered into the one-way linked list clockwise starting from the boundary pixel point. According to the total number of boundary pixel points, the identification interval of the flame tip is determined to be n boundary pixel points, and the boundary pixel points in the one-way linked list whose Euclidean distance is greater than the Euclidean distance of n / 2 boundary pixel points in the left and right neighborhoods are marked as possible flame tips;

[0067] The sum of the Euclidean distances S1, S2 and S3 of the n / 8, n / 4 and n / 2 boundary pixel points in the left and right areas of the possible flame corner are obtained respectively. When S1<S2<S3, the possible flame corner is converted into a suspected flame corner.

[0068] An approximate triangle is constructed with the two-dimensional coordinates of the suspected flame corner as the vertex and the two-dimensional coordinates of n boundary pixels in the flame corner identification interval as the base, and the tangent value of the standard vertex angle half-angle of the approximate triangle of the flame corner is set. When the tangent value of the vertex angle half-angle of the approximate triangle of the suspected flame corner is less than or equal to the tangent value of the standard vertex angle half-angle, the suspected flame corner is converted into a flame corner.

[0069] It should be further explained that, in a specific implementation process, the process of judging whether a fire occurs in the area in the video frame image based on the number of flame corners in the suspicious area and judging the type of the fire includes:

[0070] Select the number of flame corners of the video frame image at the tth second, and compare the number of flame corners of the video frame image with a preset number of flame corners;

[0071] When the number of flame sharp corners is greater than a preset number of flame sharp corners, the video frame image is marked as a fire source image, and a suspicious area in the video frame image is marked as a flame area;

[0072] When the number of flame sharp corners is less than or equal to the preset number of flame sharp corners, marking the video frame image as an image of other interference sources;

[0073] If the video frame image at the t second is a fire source image, the number of flame corners in the video frame image is averaged by difference with the number of flame corners in the video frame image at the t-1 second, the number of flame corners in the video frame image at the t-2 second, and the number of flame corners in the video frame image at the t-3 second, respectively, to obtain the average value of the change in the corners of the video frame image at the t second and three consecutive adjacent video frame images;

[0074] Set the average threshold of sharp angle change, and compare the average value of the sharp angle change of the video frame image at the tth second with the average threshold of the sharp angle change;

[0075] If the average value of the sharp angle change is greater than the average threshold value of the sharp angle change, the fire source in the video frame image is marked as an out-of-control fire source;

[0076] If the average value of the sharp angle change is less than or equal to the average threshold of the sharp angle change, the fire source in the video frame image is marked as a stable fire source.

[0077] It should be further explained that, in the specific implementation process, the process of constructing a regression model through the video frame images of the fire and obtaining the fire change trend prediction results includes:

[0078] The continuous video frame images of the fire source being the uncontrolled fire source in the fire source image are sampled at equal intervals to obtain k sample video frame images;

[0079] Generate a flame area according to the number of pixels contained in the flame area in the video frame image, and generate a centroid of the flame area according to the pixel values ​​of the pixels contained in the flame area in the video frame image;

[0080] The flame area and the flame area centroid of k sample video frame images are curve-fitted using a regression model to obtain a flame area growth curve and a flame area centroid movement curve respectively;

[0081] The flame area change trend and the flame spreading direction trend in the next video frame image are obtained according to the flame area growth curve and the flame area centroid movement curve.

[0082] It should be further explained that, in the specific implementation process, the process of visually displaying the fire change trend prediction results through the comprehensive pipe gallery full area monitoring visual map includes:

[0083] Obtain the analysis result of the video frame image of each fire video monitoring point. If the analysis result of the video frame image is a fire source image, and the fire source is an out-of-control fire source, the comprehensive pipe gallery full-area monitoring visual map will display the comprehensive pipe gallery sub-area where the out-of-control fire source is located as a red alarm area;

[0084] If the analysis result of the video frame image is a fire source image, and the fire source is a stable fire source, the comprehensive pipe gallery full-area monitoring visual map will display the comprehensive pipe gallery sub-area where the out-of-control fire source is located as an orange alarm area;

[0085] The visual graph of the comprehensive pipe gallery full-area monitoring obtains the fire spread arrival time and fire spread range of the non-alarm area according to the flame area change trend and flame spread direction trend of the red alarm area, and sets the non-alarm area within the fire spread range as the time-divided early warning area according to the fire spread arrival time.

[0086] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A fire safety early warning method for a comprehensive pipe gallery based on image processing, characterized in that: The following steps are involved: Step S1: Obtain the pipe gallery information of the current integrated pipe gallery fire management area, and set fire video monitoring points according to the pipe gallery information; Step S2: construct a visual map of the entire area of ​​the integrated pipe gallery, filter out video frame images containing suspicious areas through the video data of each fire video monitoring point, and determine the number of flame corners in the suspicious area by setting the identification interval of flame corners, the average value of the corners at different stages in the identification interval, and the approximate triangular shape of the flame corners; The process of determining the number of flame corners in a suspicious area by setting three screening steps, namely, the flame corner identification interval, the average value of the corners at different stages within the identification interval, and the approximate triangular shape of the flame corners, includes: Select the pixel point at the center of the suspicious area in the binary image as the central pixel point, establish a two-dimensional coordinate system with the central pixel point as the origin, map all the pixels in the suspicious area to the two-dimensional coordinate system, and obtain the Euclidean distance between each boundary pixel point on the edge of the suspicious area and the central pixel point; Construct a one-way linked list, randomly select a boundary pixel point on the edge of the suspicious area in the binary image, and enter the two-dimensional coordinates and Euclidean distances of all boundary pixels on the edge of the suspicious area clockwise into the one-way linked list in sequence, starting from the boundary pixel point, and determine the identification interval of the flame tip as n boundary pixels according to the total number of boundary pixels, and mark the boundary pixels in the one-way linked list whose Euclidean distance is greater than the Euclidean distance of n / 2 boundary pixels in the left and right neighborhoods as possible flame tips; The sum of the Euclidean distances S1, S2 and S3 of the n / 8, n / 4 and n / 2 boundary pixel points in the left and right neighborhoods of the possible flame corner are obtained respectively. When S1<S2<S3, the possible flame corner is converted into a suspected flame corner. An approximate triangle is constructed by using the two-dimensional coordinates of the suspected flame corner as the vertex and the two-dimensional coordinates of n boundary pixel points in the flame corner identification interval as the base, and the tangent value of the standard vertex half angle of the approximate triangle of the flame corner is set; when the tangent value of the vertex half angle of the approximate triangle of the suspected flame corner is less than or equal to the tangent value of the standard vertex half angle, the suspected flame corner is converted into a flame corner; Step S3: judging whether a fire occurs in the area in the video frame image based on the number of flame corners in the suspicious area, and judging the type of fire; Step S4: construct a regression model through the video frame images of the fire, obtain the fire change trend prediction results, and visualize the fire change trend prediction results through the comprehensive pipe gallery full-area monitoring visual diagram.

2. The method for early warning of fire safety in a comprehensive pipe gallery based on image processing according to claim 1 is characterized in that: The process of obtaining the pipe gallery information of the current integrated pipe gallery fire management area and setting the fire video monitoring points according to the pipe gallery information includes: Obtain the municipal pipeline laying characteristics of the current integrated pipe gallery fire management area, extract the pipe gallery information according to the municipal pipeline laying characteristics, split the integrated pipe gallery fire management area according to the pipe gallery information, and divide it into several integrated pipe gallery sub-areas; According to the pipeline corridor information and location characteristics of each comprehensive pipeline corridor sub-area, the evaluation index is selected, the index weight matrix of the evaluation index is set, and the membership matrix of each comprehensive pipeline corridor sub-area for the preset importance level is obtained through the fuzzy comprehensive evaluation method; The importance level of each integrated pipe gallery sub-area is obtained according to the membership matrix and the indicator weight matrix, the number of fire video monitoring points is determined according to the importance level of the integrated pipe gallery sub-area, and the distribution of fire video monitoring points in the integrated pipe gallery sub-area is determined according to the effective coverage area of ​​the fire video monitoring points and the location characteristics of the integrated pipe gallery sub-area.

3. The method for early warning of fire safety in a comprehensive pipe gallery based on image processing according to claim 2 is characterized in that: The process of constructing a visual map of the entire area of ​​the integrated pipeline corridor includes: The physical entities of the municipal pipelines in the current fire management area of ​​the integrated pipe gallery are obtained, and a map of the integrated pipe gallery is constructed according to the location characteristics and pipe gallery connection relationship between the sub-areas of the integrated pipe gallery in the current fire management area of ​​the integrated pipe gallery. The physical entities of the municipal pipelines are subjected to three-dimensional modeling and mapped to the integrated pipe gallery map to obtain a three-dimensional model; The multi-source video data of each fire video monitoring point in each integrated pipe gallery sub-area is obtained, and the multi-source video data is preprocessed in data format to obtain twin data. According to the assembly connection relationship between each fire video monitoring point and each integrated pipe gallery sub-area in the physical space, the twin data is matched with the three-dimensional model to obtain a visual monitoring diagram of the entire integrated pipe gallery area.

4. The method for early warning of fire safety in a comprehensive pipe gallery based on image processing according to claim 3 is characterized in that: The process of filtering out video frame images containing suspicious areas from the video data of each fire video monitoring point includes: Obtain video data of each fire video monitoring point, convert the obtained video data into continuous video frame images, set a reference video frame image of each fire video monitoring point, subtract the pixel values ​​of the video frame image and the reference video frame image at corresponding positions to obtain a difference value, convert the difference value into a pixel value of a binary image, and generate a binary image; Set two pixel traversal pointers and start traversal from the first pixel and the last pixel in the binary image area at the same time; Mark the pixel point area in the binary image whose pixel value is greater than the preset pixel threshold as a suspicious area; The pixel point area in the binary image whose pixel value is less than or equal to the preset pixel threshold is marked as a normal area.

5. The method for early warning of fire safety in a comprehensive pipe gallery based on image processing according to claim 4 is characterized in that: The process of judging whether a fire occurs in the area in the video frame image based on the number of flame corners in the suspicious area and judging the type of the fire includes: Select the number of flame corners of the video frame image at the tth second, and compare the number of flame corners of the video frame image with a preset number of flame corners; When the number of flame sharp corners is greater than a preset number of flame sharp corners, the video frame image is marked as a fire source image, and a suspicious area in the video frame image is marked as a flame area; When the number of flame sharp corners is less than or equal to the preset number of flame sharp corners, marking the video frame image as an image of other interference sources; If the video frame image at the t second is a fire source image, the number of flame corners in the video frame image is averaged by difference with the number of flame corners in the video frame image at the t-1 second, the number of flame corners in the video frame image at the t-2 second, and the number of flame corners in the video frame image at the t-3 second, respectively, to obtain the average value of the change in the corners of the video frame image at the t second and three consecutive adjacent video frame images; Set the average threshold of sharp angle change, and compare the average value of the sharp angle change of the video frame image at the tth second with the average threshold of the sharp angle change; If the average value of the sharp angle change is greater than the average threshold value of the sharp angle change, the fire source in the video frame image is marked as an out-of-control fire source; If the average value of the sharp angle change is less than or equal to the average threshold of the sharp angle change, the fire source in the video frame image is marked as a stable fire source.

6. The method for early warning of fire safety in a utility tunnel based on image processing according to claim 5 is characterized in that: The process of building a regression model through the video frame images of the fire and obtaining the fire change trend prediction results includes: The continuous video frame images of the fire source being the uncontrolled fire source in the fire source image are sampled at equal intervals to obtain k sample video frame images; Generate a flame area according to the number of pixels contained in the flame area in the video frame image, and generate a centroid of the flame area according to the pixel values ​​of the pixels contained in the flame area in the video frame image; The flame area and the flame area centroid of k sample video frame images are curve-fitted using a regression model to obtain a flame area growth curve and a flame area centroid movement curve respectively; The flame area change trend and the flame spreading direction trend in the next video frame image are obtained according to the flame area growth curve and the flame area centroid movement curve.

7. The method for early warning of fire safety in a utility tunnel based on image processing according to claim 6 is characterized in that: The process of visually displaying the fire change trend prediction results through the comprehensive pipe gallery full area monitoring visual map includes: Obtain the analysis result of the video frame image of each fire video monitoring point. If the analysis result of the video frame image is a fire source image, and the fire source is an out-of-control fire source, the comprehensive pipe gallery full-area monitoring visual map will display the comprehensive pipe gallery sub-area where the out-of-control fire source is located as a red alarm area; If the analysis result of the video frame image is a fire source image, and the fire source is a stable fire source, the comprehensive pipe gallery full-area monitoring visual map will display the comprehensive pipe gallery sub-area where the out-of-control fire source is located as an orange alarm area; The visual graph of the comprehensive pipe gallery full-area monitoring obtains the fire spread arrival time and fire spread range of the non-alarm area according to the flame area change trend and flame spread direction trend of the red alarm area, and sets the non-alarm area within the fire spread range as the time-divided early warning area according to the fire spread arrival time.

Images