Subway flood prevention key node generation method based on subway internal environment image recognition
By recognizing images of the subway's internal environment, key flood control nodes are generated, solving the problem of insufficient research on sandbag placement and achieving better flood control results and greater flood control redundancy.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- NANJING WATER PLANNING & DESIGNING INST
- Filing Date
- 2025-10-22
- Publication Date
- 2026-07-03
AI Technical Summary
In existing technologies, there is a lack of research on the placement of sandbags relative to the transverse ditch, which limits the potential for improving flood control effectiveness and makes it impossible to effectively cope with situations where water levels rise rapidly.
By using images of the subway's internal environment, the system obtains the characteristics of the transverse ditch, identifies the benchmark location and forms the benchmark area, analyzes the water flow trend, determines the confluence point, and allocates the number of sandbags according to the size of the confluence point to generate key flood control nodes.
This improves flood control effectiveness, allowing sandbags to be piled up at key points to create more flood control redundancy and reduce the impact of flooding on subway stations.
Smart Images
Figure CN121354016B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of flood control technology, specifically to a method for generating key nodes for subway flood control based on subway internal environment image recognition. Background Technology
[0002] Subway station entrances and exits are well-known flood control locations, so there's no need for further discussion. Subway stations have drainage ditches that channel water into connected pumps. When the water level in the pumps reaches a warning level, the pumps operate to discharge the excess water, thus achieving flood control. However, the relative positions of sandbags and drainage ditches can affect flood control capabilities. Current technology lacks research on the placement of sandbags relative to drainage ditches, leaving room for improvement in the effectiveness of flood control at key points in their formation. Summary of the Invention
[0003] To address the aforementioned technical problems, this paper provides a method for generating key nodes for subway flood control based on subway internal environment image recognition. This technical solution solves the problems mentioned in the background section.
[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows:
[0005] A method for generating key nodes for subway flood control based on subway interior environment image recognition includes:
[0006] Obtain the judgment features of at least one transverse trench, and based on the judgment features, identify at least one location where the transverse trench appears inside the subway, as a reference location;
[0007] At least one reference location is merged to obtain at least one reference area, and the flood control area of the reference area is formed based on the internal environment identification.
[0008] The terrain of the flood control area is identified to obtain the water flow trend in the flood control area. Based on the water flow trend, the confluence of the flood control area is analyzed.
[0009] The number of sandbags inside the subway station is obtained as the number of features. The sandbags inside the subway station do not include those placed at the subway station entrances and exits. The walls on both sides of the confluence are identified to obtain the feature walls of the confluence.
[0010] The number of sandbags to be allocated to the confluence is determined based on the size of the confluence, and this number is used as the target number.
[0011] Based on the number of targets and the characteristic walls, generate key flood control nodes at the confluence point.
[0012] Preferably, the determination feature for obtaining at least one transverse groove includes the following steps:
[0013] Pre-acquire at least one cross-ditch image in all subways, identify black pixels in the cross-ditch image, and summarize adjacent black pixels to obtain at least one black area, which is a drainage hole.
[0014] Randomly select the area of one of the black regions as the base area, and divide the base area by the area of the black region to obtain the scaling ratio of the black region.
[0015] The transverse trench image is scaled according to the scaling ratio of the black area to obtain the initial image;
[0016] Take at least one sampling point evenly within the range of 0 to 360 degrees, and use the distance between adjacent sampling points as the feature value;
[0017] The initial image is rotated to obtain the modified image, with the center of rotation being the center of the initial image and the rotation angle being a feature value.
[0018] The difference between the initial image and the modified image is taken as the error value of the initial image. The average of the error values of at least one initial image is taken to obtain the allowable error value.
[0019] The initial image is rotated to obtain a reference image, with the rotation center being the center of the initial image and the rotation angle being the value of the sampling point;
[0020] All reference images are aggregated and deduplicated. During deduplication, if the difference between two reference images does not exceed the allowable error value, only one reference image is retained and the other reference image is deleted. Otherwise, both reference images are retained, and at least one of the deduplicated reference images is used as the judgment feature of the transverse groove.
[0021] Preferably, identifying at least one location where a transverse trench appears inside the subway as a reference location includes the following steps:
[0022] Acquire an overall image of the subway floor, uniformly select at least one recognition point within the overall image, and translate the reference image to the feature location, such that the center of the reference image coincides with the recognition point at the feature location;
[0023] The portion of the overall image covered by the reference image at the feature location is taken as the feature part. If the difference between the reference image and the feature part does not exceed the allowable error value, the actual location of the feature part is taken as the reference location; otherwise, no processing is performed.
[0024] Preferably, merging at least one reference location to obtain at least one reference region includes the following steps:
[0025] The reference positions are numbered sequentially. If two reference positions are in contact, the numbers of both reference positions are updated and replaced with the smaller value of the two numbers. Otherwise, no processing is performed.
[0026] Once contact determination is completed for all reference positions, reference positions with the same number are merged to form a reference area.
[0027] Preferably, the process of identifying the flood control area based on the internal environment includes the following steps:
[0028] The two walls closest to the center of the reference area are designated as the first wall and the second wall, respectively. At least one first sample point is uniformly taken at the bottom of the first wall and at least one second sample point is uniformly taken at the bottom of the second wall. The number of first sample points and second sample points are the same.
[0029] The nearest subway station entrance to the baseline area is taken as the target entrance. The distance from the first sample point to the target entrance is taken as the sample value of the first sample point. The distance from the second sample point to the target entrance is taken as the sample value of the second sample point.
[0030] The first sample point is numbered according to its sample value from smallest to largest, and the second sample point is numbered according to its sample value from smallest to largest.
[0031] Three-dimensional models are created for the first and second walls. The direction vector of the tangent line of the first wall at the first sample point along the horizontal direction is taken as the first vector, and the direction vector of the tangent line of the second wall at the second sample point along the horizontal direction is taken as the second vector. Both the first and second vectors point away from the target opening.
[0032] Obtain the angle between the first vector and the second vector with the same number, and use it as the feature angle. Assign the number of the first vector to the feature angle it generates.
[0033] Set a check value. The range of the check value is the number of the first vector. If the feature angles with numbers less than the check value are all less than the feature angles with numbers greater than the check value, then the check value is used as the target check value.
[0034] The minimum value of the target verification value is taken as the preset value. The line connecting the first sample point and the second sample point with the number equal to the preset value is taken as the first line segment. The line connecting the first sample point and the second sample point with the number equal to 1 is taken as the second line segment. The area enclosed by the first line segment, the second line segment, the first wall and the second wall is taken as the initial area. The initial area excluding the area where the elevator is located is taken as the flood control area of the reference area.
[0035] Preferably, the step of identifying the terrain of the flood control area to obtain the water flow trend of the flood control area includes the following steps:
[0036] The water flow trend in the flood control area follows the direction of increasing numbering of the first sample point.
[0037] Preferably, the step of analyzing the confluence of the flood control area based on the water flow trend includes the following steps:
[0038] The location of the first line segment is designated as the confluence point of the flood control area.
[0039] Preferably, identifying the characteristic walls on both sides of the confluence outlet to obtain the characteristic wall surface of the confluence outlet includes the following steps:
[0040] The portion of the first and second walls located within the flood control area of the reference zone is designated as the characteristic wall of the confluence.
[0041] Preferably, determining the number of sandbags allocated to the confluence according to the size of the confluence includes the following steps:
[0042] The total length is obtained by summing the opening lengths of the manifolds. The distribution coefficient of the manifold is obtained by dividing the opening length of the manifold by the total length.
[0043] The number of sandbags allocated to the confluence is obtained by multiplying the allocation coefficient of the confluence by the number of features.
[0044] Preferably, generating the key flood control nodes at the confluence point based on the number of targets and characteristic wall surfaces includes the following steps:
[0045] Get the length and height of the sandbags. Multiply the length of the sandbags by the target number to get the available length.
[0046] The trajectories of the intersections of the two characteristic walls at the junction with the subway floor are respectively designated as the first trajectory and the second trajectory.
[0047] At least one first feature point is uniformly selected on the first trajectory, and at least one second feature point is uniformly selected on the second trajectory;
[0048] Form at least one point set combination, which is composed of a first feature point and a second feature point randomly combined;
[0049] The line connecting the first and second feature points in the point set combination is taken as the feature line;
[0050] The feature line divides the flood control area into two identification areas: the identification area containing the confluence outlet is designated as the non-target area, and the identification area not containing the confluence outlet is designated as the target area.
[0051] If the target region completely encompasses the reference region, then the point set combination of the target region will be generated as the target point set combination;
[0052] The result of dividing the callable length by the length of the feature line is rounded to obtain the target value. The target value is multiplied by the height of the sandbag to obtain the flood control height. The flood control height is multiplied by the area of the target area to obtain the flood control volume.
[0053] The feature line generated by combining the target point set with the largest flood control volume is taken as the key flood control node at the confluence.
[0054] Compared with the prior art, the beneficial effects of the present invention are as follows:
[0055] By identifying benchmark locations, forming flood control zones within benchmark areas, obtaining the confluence points of the flood control zones, and generating key flood control nodes at the confluence points, the width variations of channels containing cross-cutting ditches can be identified based on the analysis of wall locations. Furthermore, based on these width variations, key flood control nodes at the confluence points can be determined. This ensures that sandbags are piled up at these key flood control nodes, providing greater flood control redundancy space for the corresponding cross-cutting ditches, achieving better flood control results, effectively preventing the spread of floodwaters, and minimizing the impact of flooding on the subway station. Attached Figure Description
[0056] Figure 1 This is a flowchart illustrating the method for generating key nodes in subway flood control based on subway interior environment image recognition according to the present invention.
[0057] Figure 2 This is a flowchart illustrating the process of obtaining the judgment feature of at least one transverse groove according to the present invention.
[0058] Figure 3 This is a schematic diagram illustrating the process of identifying at least one location of a cross-shaped ditch inside a subway station as a reference location according to the present invention.
[0059] Figure 4 This is a schematic diagram of the process of merging at least one reference position to obtain at least one reference region according to the present invention.
[0060] Figure 5 This is a schematic diagram of the process of forming a reference area for flood control based on internal environment identification according to the present invention;
[0061] Figure 6 This is a schematic diagram illustrating the process of determining the number of sandbags allocated to the confluence according to the size of the confluence in this invention.
[0062] Figure 7 This is a schematic diagram of the process of generating key flood control nodes at the confluence point according to the number of targets and characteristic walls of the present invention. Detailed Implementation
[0063] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.
[0064] Reference Figure 1 As shown, the method for generating key nodes for subway flood control based on subway interior environment image recognition includes:
[0065] Obtain the judgment features of at least one transverse trench, and based on the judgment features, identify at least one location where the transverse trench appears inside the subway, as a reference location;
[0066] At least one reference location is merged to obtain at least one reference area, and the flood control area of the reference area is formed based on the internal environment identification.
[0067] The terrain of the flood control area is identified to obtain the water flow trend in the flood control area. Based on the water flow trend, the confluence of the flood control area is analyzed.
[0068] The number of sandbags inside the subway station is obtained as the number of features. The sandbags inside the subway station do not include those placed at the subway station entrances and exits. The walls on both sides of the confluence are identified to obtain the feature walls of the confluence.
[0069] The number of sandbags to be allocated to the confluence is determined based on the size of the confluence, and this number is used as the target number.
[0070] Based on the number of targets and the characteristic walls, generate key flood control nodes at the confluence point.
[0071] In this plan, the subway itself is not relevant to flood prevention; the focus is on preventing excessive flooding within the subway station to avoid high water levels that could endanger passenger safety. Placing sandbags at subway entrances is common practice and requires no further discussion. However, placing sandbags at the cross-ditch is worth considering. Cross-ditches are typically located in the first passageway after entering the subway station from the entrance, near the escalators. Water flowing into the cross-ditch will flow into pumps buried within the subway station and be discharged by their motors. Generally, the water level rise caused by typical heavy rains is slow, and the steps at the subway entrances make the water level rise more manageable. The entrances and exits are about one meter higher than the surrounding ground. Stacking sandbags at the entrances and exits of the subway station can withstand ordinary floods. If it cannot withstand the floods, it means that the water level is rising rapidly and will flow into the subway station. Currently, the sandbags in the cross-ditch are usually stacked at the cross-ditch, which results in a small space for delaying the flood with the wall. When the water level rises rapidly, even if the water pumps drain water, the water will be filled up quickly and then overflow the sandbags and spread further into the subway station. Therefore, it is necessary to set the position of the sandbags according to the cross-ditch to ensure that the space for delaying the flood with the wall is as large as possible. At the same time, it is also necessary to consider the number of sandbags, because the number of sandbags in the subway station is usually small.
[0072] Generally, the width of a channel containing a transverse ditch does not vary much, meaning the distance between the walls on both sides of the channel remains relatively constant. Therefore, the end of the channel is usually used as the location for sandbag accumulation. The end of the channel leads to the open area inside the subway station, maximizing the space for flood mitigation. However, due to variations in the geology of subway stations built in different regions, there are a few cases where the width of the transverse ditch varies significantly due to geological conditions. This could be because there are rocks on both sides of the channel that need to be avoided, or the soil strength at the top of the channel is insufficient, requiring the channel to be narrowed according to the soil conditions. In short, many factors can cause significant variations in the width of the channel, making it unsuitable to use the end of the channel as the location for sandbag accumulation. Since there may be narrower areas near the end of the channel, sandbags can be piled up higher there. Although the area for water mitigation is reduced, the height is increased significantly, resulting in a larger overall flood control space and better flood prevention. Therefore, in these cases, it is necessary to generate key flood control nodes for sandbag accumulation, followed by a series of steps to handle this.
[0073] It should be noted that subway stations usually have multiple entrances and exits.
[0074] Reference Figure 2 As shown, obtaining the judgment features of at least one transverse groove includes the following steps:
[0075] Pre-acquire at least one cross-ditch image in all subways, identify black pixels in the cross-ditch image, and summarize adjacent black pixels to obtain at least one black area, which is a drainage hole.
[0076] Randomly select the area of one of the black regions as the base area, and divide the base area by the area of the black region to obtain the scaling ratio of the black region.
[0077] The transverse trench image is scaled according to the scaling ratio of the black area to obtain the initial image;
[0078] Take at least one sampling point evenly within the range of 0 to 360 degrees, and use the distance between adjacent sampling points as the feature value;
[0079] The initial image is rotated to obtain the modified image, with the center of rotation being the center of the initial image and the rotation angle being a feature value.
[0080] The difference between the initial image and the modified image is taken as the error value of the initial image. The average of the error values of at least one initial image is taken to obtain the allowable error value.
[0081] The initial image is rotated to obtain a reference image, with the rotation center being the center of the initial image and the rotation angle being the value of the sampling point;
[0082] All reference images are aggregated and deduplicated. During deduplication, if the difference between two reference images does not exceed the allowable error value, only one reference image is retained and the other reference image is deleted. Otherwise, both reference images are retained, and at least one of the deduplicated reference images is used as the judgment feature of the transverse groove.
[0083] Different subway stations may use different specifications for their cross trenches, requiring the acquisition of images for each type. The top of the cross trench is covered with a silver-white cap with holes, through which water flows into the cross trench and is discharged into the water pump. Since there is no light source inside the cross trench, the openings are black. The identification of black pixels is achieved by acquiring the pixel value corresponding to the black area. Generally, the size of the black area in the same cross trench is consistent, but the size in the image may be different. Therefore, for ease of comparison, the image is scaled down to the same size as the black area. It is easy to see that although the image capture ensures a straight view, the image may be deflected like a clock hand. To overcome this, the image is rotated to obtain a reference image, and duplicates are removed from the reference image. Thus, any image of a cross trench will inevitably differ from a certain reference image by less than the allowable error value. Therefore, cross trench identification can be performed based on this.
[0084] Adjacent black pixels refer to two black pixels that are not connected to any other pixels.
[0085] The difference between the initial image and the modified image refers to the sum of the absolute values of the differences in pixel values at the same position in the initial image and the modified image. The difference between the remaining images is handled in a similar way.
[0086] Reference Figure 3 As shown, identifying at least one location where a transverse trench appears inside the subway, as a reference location, includes the following steps:
[0087] Acquire an overall image of the subway floor, uniformly select at least one recognition point within the overall image, and translate the reference image to the feature location, such that the center of the reference image coincides with the recognition point at the feature location;
[0088] The portion of the overall image covered by the reference image at the feature location is taken as the feature part. If the difference between the reference image and the feature part does not exceed the allowable error value, the actual location of the feature part is taken as the reference location; otherwise, no processing is performed.
[0089] The overall image and the reference image use the same scale, which is easy to do. Simply scale the overall image so that the ratio of the object in the overall image to its actual size is equal to the ratio of the object in the reference image to its actual size.
[0090] Reference Figure 4 As shown, merging at least one reference location to obtain at least one reference region includes the following steps:
[0091] The reference positions are numbered sequentially. If two reference positions are in contact, the numbers of both reference positions are updated and replaced with the smaller value of the two numbers. Otherwise, no processing is performed.
[0092] Once contact determination is completed for all reference positions, reference positions with the same number are merged to form a reference area.
[0093] Reference locations with the same number are interconnected, and therefore, they can be aggregated into a reference area. The reference area as a whole is located in the same passageway leading into the subway station, which delays the flooding at the same location.
[0094] Reference Figure 5 As shown, based on internal environment identification, the flood control area of the baseline area includes the following steps:
[0095] The two walls closest to the center of the reference area are designated as the first wall and the second wall, respectively. At least one first sample point is uniformly taken at the bottom of the first wall and at least one second sample point is uniformly taken at the bottom of the second wall. The number of first sample points and second sample points are the same.
[0096] The nearest subway station entrance to the baseline area is taken as the target entrance. The distance from the first sample point to the target entrance is taken as the sample value of the first sample point. The distance from the second sample point to the target entrance is taken as the sample value of the second sample point.
[0097] The first sample point is numbered according to its sample value from smallest to largest, and the second sample point is numbered according to its sample value from smallest to largest.
[0098] Three-dimensional models are created for the first and second walls. The direction vector of the tangent line of the first wall at the first sample point along the horizontal direction is taken as the first vector, and the direction vector of the tangent line of the second wall at the second sample point along the horizontal direction is taken as the second vector. Both the first and second vectors point away from the target opening.
[0099] Obtain the angle between the first vector and the second vector with the same number, and use it as the feature angle. Assign the number of the first vector to the feature angle it generates.
[0100] Set a check value. The range of the check value is the number of the first vector. If the feature angles with numbers less than the check value are all less than the feature angles with numbers greater than the check value, then the check value is used as the target check value.
[0101] The minimum value of the target verification value is taken as the preset value. The line connecting the first sample point and the second sample point with the number equal to the preset value is taken as the first line segment. The line connecting the first sample point and the second sample point with the number equal to 1 is taken as the second line segment. The area enclosed by the first line segment, the second line segment, the first wall and the second wall is taken as the initial area. The initial area excluding the area where the elevator is located is taken as the flood control area of the reference area.
[0102] The flood control area here can be basically identified as the passage that includes the reference area. The first wall and the second wall are the walls on both sides of the passage that includes the reference area. However, the first wall and the second wall will continue to extend in two directions. One is to extend to the nearest subway station entrance and the other is to determine the flood control area of the reference area based on the first wall and the second wall. The main reason is that the passage of the reference area will suddenly become wider at the end because it connects to the open area inside the subway station. It is easy to know that the wall after this point will have a large angle change, while the angle before this point is fluctuating. Therefore, based on this, the preset value is obtained, and thus the end of the passage of the reference area is obtained. However, it should be noted that the process of forming the end here does not use the narrowest position of the end as a screening condition.
[0103] Because the passageway connects to the elevator at the subway station entrance and exit, it is necessary to remove the initial area where the elevator is located in order to obtain the flood control area of the baseline area.
[0104] The direction vector of the tangent line along the horizontal direction of the first wall at the first sample point has two directions to choose from. Therefore, the direction needs to be further determined. Here, the direction pointing away from the target opening is chosen, that is, the direction in which the water flows into the subway station. This is because as the water flow time increases, its distance from the target opening will inevitably increase. The second vector is handled in a similar way.
[0105] To identify the terrain of a flood control area and determine the water flow patterns within that area, the following steps are involved:
[0106] The water flow trend in the flood control area follows the direction of increasing numbering of the first sample point.
[0107] Based on the water flow trend, the analysis of the confluence points of the flood control area includes the following steps:
[0108] The location of the first line segment is designated as the confluence point of the flood control area.
[0109] Identifying the walls on both sides of the junction port to obtain the characteristic walls of the junction port includes the following steps:
[0110] The portion of the first and second walls located within the flood control area of the reference zone is designated as the characteristic wall of the confluence.
[0111] Reference Figure 6 As shown, determining the number of sandbags allocated to the confluence according to its size includes the following steps:
[0112] The total length is obtained by summing the opening lengths of the manifolds. The distribution coefficient of the manifold is obtained by dividing the opening length of the manifold by the total length.
[0113] The number of sandbags allocated to the confluence is obtained by multiplying the allocation coefficient of the confluence by the number of features.
[0114] Since the number of sandbags is limited, they need to be distributed according to the size of the confluence, i.e., the length of the first segment.
[0115] Reference Figure 7 As shown, generating key flood control nodes at the confluence point, based on the number of targets and characteristic walls, includes the following steps:
[0116] Get the length and height of the sandbags. Multiply the length of the sandbags by the target number to get the available length.
[0117] The trajectories of the intersections of the two characteristic walls at the junction with the subway floor are respectively designated as the first trajectory and the second trajectory.
[0118] At least one first feature point is uniformly selected on the first trajectory, and at least one second feature point is uniformly selected on the second trajectory;
[0119] Form at least one point set combination, which is composed of a first feature point and a second feature point randomly combined;
[0120] The line connecting the first and second feature points in the point set combination is taken as the feature line;
[0121] The feature line divides the flood control area into two identification areas: the identification area containing the confluence outlet is designated as the non-target area, and the identification area not containing the confluence outlet is designated as the target area.
[0122] If the target region completely encompasses the reference region, then the point set combination of the target region will be generated as the target point set combination;
[0123] The result of dividing the callable length by the length of the feature line is rounded to obtain the target value. The target value is multiplied by the height of the sandbag to obtain the flood control height. The flood control height is multiplied by the area of the target area to obtain the flood control volume.
[0124] The feature line generated by combining the target point set with the largest flood control volume is taken as the key flood control node at the confluence.
[0125] The key flood control node must be near the confluence, but the specific location still needs to be determined. An exhaustive combination method is used to determine it, ensuring that the target area covers the baseline area. Water will accumulate in the target area, and the maximum accumulation capacity is determined by the area of the target area and the height of the sandbags. Since the confluence may not be the narrowest point, other locations are narrower and the sandbags are piled higher, so the maximum water accumulation capacity is greater. Therefore, it is necessary to identify this location as the key flood control node at the confluence.
[0126] Multiply the target value by the height of the sandbags to get the flood control height. The target value is the integer part of the result obtained by dividing the available length by the length of the feature line, denoted as N. It can be seen that the sandbags can be stacked in N layers, but cannot be stacked completely in N+1 layers. There will be a gap in the N+1th layer. Therefore, the maximum height of the water level blocked is the height of N layers of sandbags, that is, the flood control height.
[0127] Furthermore, this solution also proposes a storage medium on which a computer-readable program is stored. When the computer-readable program is invoked, it executes the aforementioned method for generating key nodes for subway flood control based on subway internal environment image recognition.
[0128] It is understandable that the storage medium can be a magnetic medium, such as a floppy disk, hard disk, or magnetic tape; an optical medium, such as a DVD; or a semiconductor medium, such as a solid-state drive (SSD).
[0129] In summary, the advantages of this invention are as follows: by identifying the reference location, forming the flood control area of the reference region, obtaining the confluence of the flood control area, and generating the key flood control nodes at the confluence, the width variation of the channel containing the cross-ditch can be identified based on the point analysis of the wall surface. Then, based on the width variation of the channel, the key flood control nodes at the confluence can be determined, ensuring that sandbags are piled up at the key flood control nodes. This provides a larger flood control redundancy space for the corresponding cross-ditch, achieving better flood control results, effectively preventing the spread of floods, and minimizing the impact of floods on the subway station.
[0130] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.
Claims
1. A method for generating key nodes in subway flood control based on subway interior environment image recognition, characterized in that, include: Obtain the judgment features of at least one transverse trench, and based on the judgment features, identify at least one location where the transverse trench appears inside the subway, as a reference location; At least one reference location is merged to obtain at least one reference area, and the flood control area of the reference area is formed based on the internal environment identification. The terrain of the flood control area is identified to obtain the water flow trend in the flood control area. Based on the water flow trend, the confluence of the flood control area is analyzed. The number of sandbags inside the subway station is obtained as the number of features. The sandbags inside the subway station do not include those placed at the subway station entrances and exits. The walls on both sides of the confluence are identified to obtain the feature walls of the confluence. The number of sandbags allocated to the confluence is determined based on the size of the confluence, and this number serves as the target number. Based on the number of targets and characteristic walls, generate key flood control nodes at the confluence point; The process of generating key flood control nodes at the confluence point based on the number of targets and characteristic walls includes the following steps: Get the length and height of the sandbags. Multiply the length of the sandbags by the target number to get the available length. The trajectories of the intersections of the two characteristic walls at the junction with the subway floor are respectively designated as the first trajectory and the second trajectory. At least one first feature point is uniformly selected on the first trajectory, and at least one second feature point is uniformly selected on the second trajectory; Form at least one point set combination, which is composed of a first feature point and a second feature point randomly combined; The line connecting the first and second feature points in the point set combination is taken as the feature line; The feature line divides the flood control area into two identification areas: the identification area containing the confluence outlet is designated as the non-target area, and the identification area not containing the confluence outlet is designated as the target area. If the target region completely encompasses the reference region, then the point set combination of the target region will be generated as the target point set combination; The result of dividing the callable length by the length of the feature line is rounded to obtain the target value. The target value is multiplied by the height of the sandbag to obtain the flood control height. The flood control height is multiplied by the area of the target area to obtain the flood control volume. The feature line generated by combining the target point set with the largest flood control volume is taken as the key flood control node at the confluence.
2. The method for generating key nodes for subway flood control based on subway interior environment image recognition according to claim 1, characterized in that, The determination feature for obtaining at least one transverse groove includes the following steps: Pre-acquire at least one cross-ditch image in all subway lines, identify black pixels in the cross-ditch image, and summarize adjacent black pixels to obtain at least one black area, which is a drainage hole. Randomly select the area of one of the black regions as the base area, and divide the base area by the area of the black region to obtain the scaling ratio of the black region. The transverse trench image is scaled according to the scaling ratio of the black area to obtain the initial image; Take at least one sampling point evenly within the range of 0 to 360 degrees, and use the distance between adjacent sampling points as the feature value; The initial image is rotated to obtain the modified image, with the center of rotation being the center of the initial image and the rotation angle being a feature value. The difference between the initial image and the modified image is taken as the error value of the initial image. The average of the error values of at least one initial image is taken to obtain the allowable error value. The initial image is rotated to obtain a reference image, with the rotation center being the center of the initial image and the rotation angle being the value of the sampling point; All reference images are aggregated and deduplicated. During deduplication, if the difference between two reference images does not exceed the allowable error value, only one reference image is retained and the other reference image is deleted. Otherwise, both reference images are retained, and at least one of the deduplicated reference images is used as the judgment feature of the transverse groove.
3. The method for generating key nodes for subway flood control based on subway interior environment image recognition according to claim 2, characterized in that, The process of identifying at least one location where a transverse trench appears inside the subway, as a reference location, includes the following steps: Acquire an overall image of the subway floor, uniformly select at least one recognition point within the overall image, and translate the reference image to the feature location, such that the center of the reference image coincides with the recognition point at the feature location; The portion of the overall image covered by the reference image at the feature location is taken as the feature part. If the difference between the reference image and the feature part does not exceed the allowable error value, the actual location of the feature part is taken as the reference location; otherwise, no processing is performed.
4. The method for generating key nodes for subway flood control based on subway interior environment image recognition according to claim 3, characterized in that, The process of merging at least one reference location to obtain at least one reference region includes the following steps: The reference positions are numbered sequentially. If two reference positions are in contact, the numbers of both reference positions are updated and replaced with the smaller value of the two numbers. Otherwise, no processing is performed. Once contact determination is completed for all reference positions, reference positions with the same number are merged to form a reference area.
5. The method for generating key nodes for subway flood control based on subway interior environment image recognition according to claim 4, characterized in that, The process of identifying and forming a baseline area for flood control based on the internal environment includes the following steps: The two walls closest to the center of the reference area are designated as the first wall and the second wall, respectively. At least one first sample point is uniformly taken at the bottom of the first wall and at least one second sample point is uniformly taken at the bottom of the second wall. The number of first sample points and second sample points are the same. The nearest subway station entrance to the baseline area is taken as the target entrance. The distance from the first sample point to the target entrance is taken as the sample value of the first sample point. The distance from the second sample point to the target entrance is taken as the sample value of the second sample point. The first sample point is numbered according to its sample value from smallest to largest, and the second sample point is numbered according to its sample value from smallest to largest. Three-dimensional models are created for the first and second walls. The direction vector of the tangent line of the first wall at the first sample point along the horizontal direction is taken as the first vector, and the direction vector of the tangent line of the second wall at the second sample point along the horizontal direction is taken as the second vector. Both the first and second vectors point away from the target opening. Obtain the angle between the first vector and the second vector with the same number, and use it as the feature angle. Assign the number of the first vector to the feature angle it generates. Set a check value. The range of the check value is the number of the first vector. If the feature angles with numbers less than the check value are all less than the feature angles with numbers greater than the check value, then the check value is used as the target check value. The minimum value of the target verification value is taken as the preset value. The line connecting the first sample point and the second sample point with the number equal to the preset value is taken as the first line segment. The line connecting the first sample point and the second sample point with the number equal to 1 is taken as the second line segment. The area enclosed by the first line segment, the second line segment, the first wall and the second wall is taken as the initial area. The initial area excluding the area where the elevator is located is taken as the flood control area of the reference area.
6. The method for generating key nodes for subway flood control based on subway interior environment image recognition according to claim 5, characterized in that, The process of identifying the terrain of the flood control area and obtaining the water flow trend of the flood control area includes the following steps: The water flow trend in the flood control area follows the direction of increasing numbering of the first sample point.
7. The method for generating key nodes for subway flood control based on subway interior environment image recognition according to claim 6, characterized in that, The process of analyzing the confluence of watercourses in the flood control area based on water flow trends includes the following steps: The location of the first line segment is designated as the confluence point of the flood control area.
8. The method for generating key nodes for subway flood control based on subway interior environment image recognition according to claim 7, characterized in that, Identifying the walls on both sides of the confluence outlet to obtain the characteristic walls of the confluence outlet includes the following steps: The portion of the first and second walls located within the flood control area of the reference zone is designated as the characteristic wall of the confluence.
9. The method for generating key nodes for subway flood control based on subway interior environment image recognition according to claim 8, characterized in that, Determining the number of sandbags allocated to the confluence according to its size includes the following steps: The total length is obtained by summing the opening lengths of the manifolds. The distribution coefficient of the manifold is obtained by dividing the opening length of the manifold by the total length. The number of sandbags allocated to the confluence is obtained by multiplying the allocation coefficient of the confluence by the number of features.