A bridge crack detection system and method based on machine vision

Abstract

The present invention provides a bridge crack detection system and method based on machine vision, wherein the system includes: a crack detection map acquisition module, used to acquire a preset crack detection map corresponding to a target bridge; an image acquisition route planning module, used to plan an image acquisition route based on the crack detection map; an image acquisition control module, used to control a drone to acquire images of the target bridge based on the image acquisition route and acquire acquired images; a crack detection module, used to perform crack detection based on the acquired images based on machine vision, and output the detection results. The present invention controls the drone to acquire images of the target bridge, and performs crack detection based on the acquired images acquired by the drone based on machine vision, thereby reducing labor costs and avoiding the problem that manual inspections have a certain degree of risk due to the passage of many vehicles on the bridge. The drone can fly to take pictures of places that cannot be reached by staff, such as the bottom of the bridge and bridge piers, thereby improving convenience.

Description

Technical Field

[0001] The present invention relates to the technical field of computer data processing, and in particular to a bridge crack detection system and method based on machine vision. Background Art

[0002] At present, cracks may appear on bridges after they have been in use for a long time. For example, the concrete surface of the bridge body may be rough, causing rainwater to enter through the rough surface and contact the internal steel bars, causing the steel bars to rust and expand, resulting in concrete cracks, etc. In order to ensure the safety of the bridge, it is necessary to regularly detect cracks on the bridge.

[0003] Traditionally, crack inspections on bridges are mostly done manually, that is, multiple staff members are arranged to inspect the bridge to see if there are cracks on the bridge. However, this has several disadvantages: first, the labor cost is high; second, there are many vehicles passing on the bridge, so manual inspections have a certain degree of risk; third, it is difficult for staff to inspect places that cannot be reached, such as the bottom of the bridge and the piers.

[0004] Therefore, a solution is needed urgently to address the above-mentioned shortcomings. Summary of the invention

[0005] One of the purposes of the present invention is to provide a bridge crack detection system based on machine vision, which controls a drone to collect images of a target bridge. Based on machine vision, crack detection is performed according to the images collected by the drone. There is no need to arrange multiple staff members to patrol the bridge to check whether cracks appear on the bridge body, thereby reducing labor costs and avoiding the problem that manual inspections have a certain degree of risk due to the passage of many vehicles on the bridge. The drone can fly to take pictures of places such as the bottom of the bridge and bridge piers that staff cannot reach, thereby improving the convenience of bridge crack detection.

[0006] An embodiment of the present invention provides a bridge crack detection system based on machine vision, comprising:

[0007] A crack detection map acquisition module is used to acquire a preset crack detection map corresponding to a target bridge;

[0008] An image acquisition route planning module is used to plan an image acquisition route based on a crack detection map;

[0009] An image acquisition control module is used to control the UAV to acquire images of the target bridge based on the image acquisition route and obtain the acquired images;

[0010] The crack detection module is used to perform crack detection based on machine vision and collected images, and output the detection results.

[0011] Preferably, the image acquisition route planning module plans the image acquisition route based on the crack detection map and performs the following operations:

[0012] Obtain detection priority values ​​of multiple detection areas in a crack detection map;

[0013] Based on the minimum bounding sphere drawing condition, a minimum bounding sphere including at least two detection areas is drawn in the crack detection map;

[0014] Get the shooting position and shooting angle corresponding to the minimum enclosing sphere;

[0015] Based on the detection order value determination rule, the detection order value is determined according to the detection priority value of the detection area in the minimum enclosing sphere;

[0016] Sort the corresponding shooting positions according to the detection order value from large to small to obtain a position sequence;

[0017] Plan flight routes between two shooting positions in the position sequence in the crack detection map;

[0018] Connect the flight route head to tail and mark the shooting angle at the corresponding shooting position on the flight route to obtain the image acquisition route;

[0019] Among them, the minimum bounding sphere drawing conditions include:

[0020] The direction angles between the two opposite directions of the surface of the detection area in the minimum enclosing sphere all fall within the preset angle range;

[0021] The minimum distances between any two of the regional surfaces are all less than or equal to a preset first distance threshold;

[0022] On the surface of each region, there is a target straight line parallel to the same diameter of the minimum enclosing sphere; both ends of the target straight line fall on the boundary line of the region surface, and the minimum distance between the surface center position of the region surface and the target straight line is less than or equal to a preset second distance threshold;

[0023] The volume of the minimum bounding sphere is less than or equal to the preset volume threshold;

[0024] The first differences between the detection priority values ​​of the detection areas in the minimum enclosing sphere are all less than or equal to a preset first difference threshold;

[0025] Among them, the detection order value determination rules include:

[0026] Determine a maximum priority value and a minimum priority value among the detection priority values ​​of the detection area in the minimum enclosing sphere;

[0027] When the second difference between the maximum priority value and the minimum priority value is greater than or equal to the preset second difference threshold, the maximum priority value is used as the detection order value; otherwise, the average value of the detection priority values ​​of the detection area in the minimum enclosing sphere is used as the detection order value.

[0028] Preferably, the image acquisition route planning module obtains the detection priority values ​​of multiple detection areas in the crack detection map and performs the following operations:

[0029] Determine the region information of the detection region from the crack detection map;

[0030] Based on a preset first feature extraction template, feature extraction is performed on the regional information to obtain an information feature set;

[0031] Obtaining a preset detection priority value determination library; the detection priority value determination library includes: multiple sets of one-to-one corresponding standard information feature sets and standard values;

[0032] determining whether the information feature set includes any standard information feature set;

[0033] When it is yes, the standard value corresponding to the standard information feature set included in the information feature set is used as the target value;

[0034] Each target value is cumulatively calculated to obtain the detection priority value of the detection area.

[0035] Preferably, the image acquisition route planning module obtains the shooting position and shooting angle corresponding to the minimum enclosing sphere and performs the following operations:

[0036] Constructing a first direction vector based on the surface center position and the surface facing direction of the region surface of the detection region in the minimum bounding sphere;

[0037] Set the preset virtual drone corresponding to the drone next to the minimum enclosing sphere;

[0038] Continuously obtain the flight position and shooting direction of the virtual drone;

[0039] Constructing a second direction vector based on the flight position and the shooting direction;

[0040] The virtual drone is flight-adjusted; when the virtual drone is flight-adjusted, the vector angle between any first direction vector and the second direction vector falls within a preset vector angle range;

[0041] Until the maximum shooting range of the virtual drone just includes the surface of the entire detection area in the minimum enclosing sphere, the current flight position of the virtual drone is used as the shooting position corresponding to the minimum enclosing sphere, and the current shooting direction of the virtual drone is used as the shooting angle corresponding to the minimum enclosing sphere.

[0042] Preferably, the crack detection module is based on machine vision and performs crack detection according to the collected images, and performs the following operations:

[0043] Based on a preset second feature extraction template, feature extraction is performed on the collected image to obtain multiple image features;

[0044] Matching the image features with standard image features in a preset standard image feature library;

[0045] When the matching is successful, it is determined that a bridge crack appears on the corresponding acquired image.

[0046] An embodiment of the present invention provides a bridge crack detection method based on machine vision, comprising:

[0047] Step S1: obtaining a preset crack detection map corresponding to a target bridge;

[0048] Step S2: planning an image acquisition route based on the crack detection map;

[0049] Step S3: Based on the image acquisition route, the UAV is controlled to acquire images of the target bridge to obtain the acquired images;

[0050] Step S4: Based on machine vision, crack detection is performed according to the collected images, and the detection results are output.

[0051] Preferably, step S2: planning an image acquisition route based on the crack detection map includes:

[0052] Obtain detection priority values ​​of multiple detection areas in a crack detection map;

[0053] Based on the minimum bounding sphere drawing condition, a minimum bounding sphere including at least two detection areas is drawn in the crack detection map;

[0054] Get the shooting position and shooting angle corresponding to the minimum enclosing sphere;

[0055] Based on the detection order value determination rule, the detection order value is determined according to the detection priority value of the detection area in the minimum enclosing sphere;

[0056] Sort the corresponding shooting positions according to the detection order value from large to small to obtain a position sequence;

[0057] Plan flight routes between two shooting positions in the position sequence in the crack detection map;

[0058] Connect the flight route head to tail and mark the shooting angle at the corresponding shooting position on the flight route to obtain the image acquisition route;

[0059] Among them, the minimum bounding sphere drawing conditions include:

[0060] The direction angles between the two opposite directions of the surface of the detection area in the minimum enclosing sphere all fall within the preset angle range;

[0061] The minimum distances between any two of the regional surfaces are all less than or equal to a preset first distance threshold;

[0062] On the surface of each region, there is a target straight line parallel to the same diameter of the minimum enclosing sphere; both ends of the target straight line fall on the boundary line of the region surface, and the minimum distance between the surface center position of the region surface and the target straight line is less than or equal to a preset second distance threshold;

[0063] The volume of the minimum bounding sphere is less than or equal to the preset volume threshold;

[0064] The first differences between the detection priority values ​​of the detection areas in the minimum enclosing sphere are all less than or equal to a preset first difference threshold;

[0065] Among them, the detection order value determination rules include:

[0066] Determine a maximum priority value and a minimum priority value among the detection priority values ​​of the detection area in the minimum enclosing sphere;

[0067] When the second difference between the maximum priority value and the minimum priority value is greater than or equal to the preset second difference threshold, the maximum priority value is used as the detection order value; otherwise, the average value of the detection priority values ​​of the detection area in the minimum enclosing sphere is used as the detection order value.

[0068] Preferably, obtaining the detection priority values ​​of multiple detection areas in the crack detection map includes:

[0069] Determine the region information of the detection region from the crack detection map;

[0070] Based on a preset first feature extraction template, feature extraction is performed on the regional information to obtain an information feature set;

[0071] Obtaining a preset detection priority value determination library; the detection priority value determination library includes: multiple sets of one-to-one corresponding standard information feature sets and standard values;

[0072] determining whether the information feature set includes any standard information feature set;

[0073] When it is yes, the standard value corresponding to the standard information feature set included in the information feature set is used as the target value;

[0074] Each target value is cumulatively calculated to obtain the detection priority value of the detection area.

[0075] Preferably, obtaining the shooting position and shooting angle corresponding to the minimum enclosing sphere includes:

[0076] Constructing a first direction vector based on the surface center position and the surface facing direction of the region surface of the detection region in the minimum bounding sphere;

[0077] Set the preset virtual drone corresponding to the drone next to the minimum enclosing sphere;

[0078] Continuously obtain the flight position and shooting direction of the virtual drone;

[0079] Constructing a second direction vector based on the flight position and the shooting direction;

[0080] The virtual drone is flight-adjusted; when the virtual drone is flight-adjusted, the vector angle between any first direction vector and the second direction vector falls within a preset vector angle range;

[0081] Until the maximum shooting range of the virtual drone just includes the surface of the entire detection area in the minimum enclosing sphere, the current flight position of the virtual drone is used as the shooting position corresponding to the minimum enclosing sphere, and the current shooting direction of the virtual drone is used as the shooting angle corresponding to the minimum enclosing sphere.

[0082] Preferably, step S4: performing crack detection based on machine vision and collected images includes:

[0083] Based on a preset second feature extraction template, feature extraction is performed on the collected image to obtain multiple image features;

[0084] Matching the image features with standard image features in a preset standard image feature library;

[0085] When the matching is successful, it is determined that a bridge crack appears on the corresponding acquired image.

[0086] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings.

[0087] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0088] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0089] Figure 1 Schematic diagram of a bridge crack detection system based on machine vision in an embodiment of the present invention;

[0090] Figure 2 Schematic diagram of a bridge crack detection method based on machine vision in an embodiment of the present invention. DETAILED DESCRIPTION

[0091] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0092] The embodiment of the present invention provides a bridge crack detection system based on machine vision, such as Figure 1 As shown, including:

[0093] The crack detection map acquisition module 1 is used to acquire a preset crack detection map corresponding to a target bridge; the target bridge is a bridge that needs to be crack-detected; and the crack detection map is marked with a plurality of detection areas on the target bridge that need to be crack-detected;

[0094] The image acquisition route planning module 2 is used to plan the image acquisition route based on the crack detection map; the image acquisition route passes through each detection area;

[0095] The image acquisition control module 3 is used to control the UAV to acquire images of the target bridge based on the image acquisition route, and obtain the acquired images; each time the UAV arrives at a detection area, it takes a photo of the detection area and transmits the captured images back;

[0096] The crack detection module 4 is used to perform crack detection based on the collected images based on machine vision and output the detection results. The system detects whether cracks appear on the bridge based on the collected images and reports to the management personnel if cracks appear.

[0097] The working principle and beneficial effects of the above technical solution are:

[0098] This application controls a drone to collect images of a target bridge, and performs crack detection based on the images collected by the drone based on machine vision. There is no need to arrange multiple staff members to patrol the bridge to check whether there are cracks in the bridge body, which reduces labor costs and avoids the problem that manual inspections have a certain degree of risk due to the passage of many vehicles on the bridge. The drone can fly to take pictures of places such as the bottom of the bridge and piers that staff cannot reach, thereby improving the convenience of bridge crack detection.

[0099] In one embodiment, the image acquisition route planning module 2 plans the image acquisition route based on the crack detection map and performs the following operations:

[0100] Obtain detection priority values ​​of multiple detection areas in the crack detection map; the larger the detection priority value, the more the corresponding detection area needs to be prioritized for image acquisition, thereby giving priority to crack detection;

[0101] Based on the minimum bounding sphere drawing condition, a minimum bounding sphere including at least two detection areas is drawn in the crack detection map;

[0102] Get the shooting position and shooting angle corresponding to the minimum enclosing sphere;

[0103] Based on the detection order value determination rule, the detection order value is determined according to the detection priority value of the detection area in the minimum enclosing sphere;

[0104] Sort the corresponding shooting positions according to the detection order value from large to small to obtain a position sequence;

[0105] Plan flight routes between two shooting positions in the position sequence in the crack detection map;

[0106] The flight route is connected end to end, and the shooting angle is marked at the corresponding shooting position on the flight route to obtain the image acquisition route; therefore, when the drone is controlled based on the image acquisition route, the drone is controlled to fly along the image acquisition route, and when it arrives at the shooting position, the drone is controlled to stay, and the image of the detection area within the minimum enclosing sphere is acquired at the shooting angle marked at the shooting position;

[0107] Among them, the minimum bounding sphere drawing conditions include:

[0108] The angles between the two facing directions of the surface of the detection area in the minimum enclosing sphere all fall within the preset angle range; the facing direction is the direction perpendicular to the center of the surface of the area; the angle range can be, for example, 0 to 20 degrees; when this condition is met, the facing directions of the detection areas in the minimum enclosing sphere are roughly the same, so that the drone can capture all the detection areas in the minimum enclosing sphere at an angle when shooting;

[0109] The minimum distance between any two of the surface areas is less than or equal to a preset first distance threshold; the first distance threshold may be, for example, 0.8 meters; when this condition is met, the detection areas in the minimum enclosing sphere will not be too far apart, thereby improving compactness;

[0110] There is a target straight line on the surface of each area that is parallel to the same diameter of the minimum enclosing sphere; when this condition is met, the surfaces of each area can be made approximately parallel, so that when the drone is shooting, it can have an angle that can approximately face the entire detection area in the minimum enclosing sphere;

[0111] The two ends of the target straight line fall on the boundary line of the regional surface, and the minimum distance between the surface center position of the regional surface and the target straight line is less than or equal to a preset second distance threshold; the second distance threshold may be, for example, 0.4 meters; when this condition is met, the larger the span of the target straight line across the regional surface, the more approximately parallel the regional surfaces are;

[0112] The volume of the minimum enclosing sphere is less than or equal to a preset volume threshold; the volume threshold may be, for example, 10 cubic meters; when this condition is met, the drone can shoot the entire detection area in the minimum enclosing sphere without causing unclear shooting due to the need to shoot too large a picture;

[0113] The first differences between the detection priority values ​​of the detection areas in the minimum enclosing sphere are all less than or equal to the preset first difference threshold; the first difference threshold may be, for example, 10; when this condition is met, the detection areas in the minimum enclosing sphere may have roughly the same degree of priority for image acquisition, thereby avoiding wasting image acquisition resources on detection areas with a lower degree of priority for image acquisition when the drone captures all the detection areas in the minimum enclosing sphere;

[0114] Among them, the detection order value determination rules include:

[0115] Determine a maximum priority value and a minimum priority value among the detection priority values ​​of the detection area in the minimum enclosing sphere;

[0116] When the second difference between the maximum priority value and the minimum priority value is greater than or equal to the preset second difference threshold, the maximum priority value is used as the detection order value; otherwise, the average value of the detection priority values ​​of the detection area in the minimum enclosing sphere is used as the detection order value. The second difference threshold can be, for example, 15; when the second difference is greater than or equal to the second difference threshold, it means that the degree to which the detection areas in the minimum enclosing sphere need to prioritize image acquisition is quite different, but the detection area corresponding to the maximum priority value itself needs to prioritize image acquisition to a greater extent, and image acquisition needs to be performed as soon as possible, so the maximum priority value is used as the detection order value, otherwise, the average value of the detection priority values ​​of the detection areas in the minimum enclosing sphere is used as the detection order value, thereby improving the rationality of the determination of the detection order value.

[0117] The working principle and beneficial effects of the above technical solution are:

[0118] Based on the crack detection map, when planning the image acquisition route, the detection order value, the minimum enclosing sphere, the shooting position and the shooting angle are introduced to improve the rationality and planning efficiency of the image acquisition route planning. Generally, the shooting range of the drone is not very small, and a large number of detection areas can be photographed at one time. If the drone is controlled to collect images of the detection areas one by one, it will cause a waste of shooting resources and reduce the shooting efficiency of the drone. Therefore, the embodiment of the present invention introduces the minimum enclosing sphere and controls the drone to collect images of all detection areas within the minimum enclosing sphere, which is particularly applicable.

[0119] In one embodiment, the image acquisition route planning module 2 obtains the detection priority values ​​of multiple detection areas in the crack detection map and performs the following operations:

[0120] Determine the regional information of the detection area from the crack detection map; the regional information includes: bridge structure information of the detection area (for example, whether there are steel bars inside), historical inspection information (for example, whether there have been pitting on the concrete surface of a certain part of the bridge in history, etc.);

[0121] Based on the preset first feature extraction template, feature extraction is performed on the regional information to obtain an information feature set; the information feature set includes: bridge structure features (such as steel bars and concrete) and historical inspection features (such as pitting on the concrete surface);

[0122] Obtaining a preset detection priority value determination library; the detection priority value determination library includes: multiple sets of one-to-one corresponding standard information feature sets and standard values;

[0123] Determine whether the information feature set contains any standard information feature set; if it does, all features in the standard information feature set are present in the information feature set;

[0124] When it is yes, the standard value corresponding to the standard information feature set included in the information feature set is used as the target value; when the information feature set includes the standard information feature set, the corresponding standard value can reflect the degree to which the detection area needs to be prioritized for image acquisition. For example, if the standard information feature set is steel bars, concrete, and pitting on the concrete surface, it means that the steel bars inside the detection area may be soaked by rainwater through the pitting surface, causing rust expansion and cracks, and the standard value is 50;

[0125] Each target value is cumulatively calculated to obtain the detection priority value of the detection area.

[0126] The working principle and beneficial effects of the above technical solution are:

[0127] The introduction of the detection priority value determination library has improved the efficiency and accuracy of determining the detection priority values ​​of the detection area.

[0128] In one embodiment, the image acquisition route planning module 2 obtains the shooting position and shooting angle corresponding to the minimum bounding sphere and performs the following operations:

[0129] Constructing a first direction vector based on the surface center position and the surface facing direction of the region surface of the detection region in the minimum bounding sphere;

[0130] The preset virtual drone corresponding to the drone is set next to the minimum enclosing sphere; the virtual drone is a drone model of a 1:1 drone;

[0131] Continuously obtain the flight position and shooting direction of the virtual drone;

[0132] Constructing a second direction vector based on the flight position and the shooting direction;

[0133] The virtual drone is flight-adjusted; when the virtual drone is flight-adjusted, the vector angle between any first direction vector and the second direction vector falls within a preset vector angle interval; the vector angle interval may be, for example, 150 degrees to 180 degrees; when the vector angle falls within the vector angle interval, it indicates that the shooting angle of the drone is approximately facing the entire detection area within the minimum enclosing sphere;

[0134] Until the maximum shooting range of the virtual drone just includes the surface of the entire detection area in the minimum enclosing sphere, the current flight position of the virtual drone is used as the shooting position corresponding to the minimum enclosing sphere, and the current shooting direction of the virtual drone is used as the shooting angle corresponding to the minimum enclosing sphere. Generally, when the virtual drone is set next to the minimum enclosing sphere, it is relatively close to the minimum enclosing sphere. At this time, the virtual drone cannot capture all the detection areas in the minimum enclosing sphere. Therefore, it is necessary to adjust the flight of the virtual drone. However, if it is adjusted randomly until the virtual drone captures all the detection areas in the minimum enclosing sphere, it will take a long time. The adjustment has no direction and lacks directionality. Therefore, the embodiment of the present invention sets the adjustment direction, that is, the vector angle between any first direction vector and the second direction vector falls within the preset vector angle interval, so as to improve the efficiency of position adjustment of the virtual drone.

[0135] In one embodiment, the crack detection module 4 performs crack detection based on the collected images based on machine vision and performs the following operations:

[0136] Based on a preset second feature extraction template, feature extraction is performed on the collected image to obtain multiple image features; the image features include: multiple local images segmented from the collected image, etc.;

[0137] Matching the image features with standard image features in a preset standard image feature library; the standard image features are bridge crack images;

[0138] When the match is met, it is determined that a bridge crack appears on the corresponding acquired image. Once the image features match the standard image features, it means that a crack exists.

[0139] The embodiment of the present invention provides a bridge crack detection method based on machine vision, such as Figure 2 As shown, including:

[0140] Step S1: obtaining a preset crack detection map corresponding to a target bridge;

[0141] Step S2: planning an image acquisition route based on the crack detection map;

[0142] Step S3: Based on the image acquisition route, the UAV is controlled to acquire images of the target bridge to obtain the acquired images;

[0143] Step S4: Based on machine vision, crack detection is performed according to the collected images, and the detection results are output.

[0144] Step S2: Based on the crack detection map, plan the image acquisition route, including:

[0145] Obtain detection priority values ​​of multiple detection areas in a crack detection map;

[0146] Based on the minimum bounding sphere drawing condition, a minimum bounding sphere including at least two detection areas is drawn in the crack detection map;

[0147] Get the shooting position and shooting angle corresponding to the minimum enclosing sphere;

[0148] Based on the detection order value determination rule, the detection order value is determined according to the detection priority value of the detection area in the minimum enclosing sphere;

[0149] Sort the corresponding shooting positions according to the detection order value from large to small to obtain a position sequence;

[0150] Plan flight routes between two shooting positions in the position sequence in the crack detection map;

[0151] Connect the flight route head to tail and mark the shooting angle at the corresponding shooting position on the flight route to obtain the image acquisition route;

[0152] Among them, the minimum bounding sphere drawing conditions include:

[0153] The direction angles between the two opposite directions of the surface of the detection area in the minimum enclosing sphere all fall within the preset angle range;

[0154] The minimum distances between any two of the regional surfaces are all less than or equal to a preset first distance threshold;

[0155] On the surface of each region, there is a target straight line parallel to the same diameter of the minimum enclosing sphere; both ends of the target straight line fall on the boundary line of the region surface, and the minimum distance between the surface center position of the region surface and the target straight line is less than or equal to a preset second distance threshold;

[0156] The volume of the minimum bounding sphere is less than or equal to the preset volume threshold;

[0157] The first differences between the detection priority values ​​of the detection areas in the minimum enclosing sphere are all less than or equal to a preset first difference threshold;

[0158] Among them, the detection order value determination rules include:

[0159] Determine a maximum priority value and a minimum priority value among the detection priority values ​​of the detection area in the minimum enclosing sphere;

[0160] When the second difference between the maximum priority value and the minimum priority value is greater than or equal to the preset second difference threshold, the maximum priority value is used as the detection order value; otherwise, the average value of the detection priority values ​​of the detection area in the minimum enclosing sphere is used as the detection order value.

[0161] Get the detection priority values ​​of multiple detection areas in the crack detection map, including:

[0162] Determine the region information of the detection region from the crack detection map;

[0163] Based on a preset first feature extraction template, feature extraction is performed on the regional information to obtain an information feature set;

[0164] Obtaining a preset detection priority value determination library; the detection priority value determination library includes: multiple sets of one-to-one corresponding standard information feature sets and standard values;

[0165] determining whether the information feature set includes any standard information feature set;

[0166] When it is yes, the standard value corresponding to the standard information feature set included in the information feature set is used as the target value;

[0167] Each target value is cumulatively calculated to obtain the detection priority value of the detection area.

[0168] Get the shooting position and shooting angle corresponding to the minimum bounding sphere, including:

[0169] Constructing a first direction vector based on the surface center position and the surface facing direction of the region surface of the detection region in the minimum bounding sphere;

[0170] Set the preset virtual drone corresponding to the drone next to the minimum enclosing sphere;

[0171] Continuously obtain the flight position and shooting direction of the virtual drone;

[0172] Constructing a second direction vector based on the flight position and the shooting direction;

[0173] The virtual drone is flight-adjusted; when the virtual drone is flight-adjusted, the vector angle between any first direction vector and the second direction vector falls within a preset vector angle range;

[0174] Until the maximum shooting range of the virtual drone just includes the surface of the entire detection area in the minimum enclosing sphere, the current flight position of the virtual drone is used as the shooting position corresponding to the minimum enclosing sphere, and the current shooting direction of the virtual drone is used as the shooting angle corresponding to the minimum enclosing sphere.

[0175] Step S4: Based on machine vision, crack detection is performed according to the collected images, including:

[0176] Based on a preset second feature extraction template, feature extraction is performed on the collected image to obtain multiple image features;

[0177] Matching the image features with standard image features in a preset standard image feature library;

[0178] When the matching is successful, it is determined that a bridge crack appears on the corresponding acquired image.

[0179] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A bridge crack detection system based on machine vision, characterized in that: include: A crack detection map acquisition module is used to acquire a preset crack detection map corresponding to a target bridge; An image acquisition route planning module, used for planning an image acquisition route based on the crack detection map; An image acquisition control module, used to control the UAV to acquire images of the target bridge based on the image acquisition route, and obtain the acquired images; A crack detection module, used to perform crack detection based on the collected image based on machine vision and output the detection result; The image acquisition route planning module plans an image acquisition route based on the crack detection map and performs the following operations: Obtaining detection priority values ​​of multiple detection areas in the crack detection map; Based on a minimum bounding sphere drawing condition, drawing a minimum bounding sphere including at least two of the detection areas in the crack detection map; Obtaining a shooting position and a shooting angle corresponding to the minimum enclosing sphere; Based on a detection order value determination rule, determining a detection order value according to the detection priority value of the detection area in the minimum enclosing sphere; The corresponding shooting positions are sorted from large to small according to the detection order values ​​to obtain a position sequence; Planning a flight route between each of the shooting positions in the position sequence in the crack detection map; Connecting the flight routes end to end and marking the shooting angles at the corresponding shooting positions on the flight routes to obtain the image acquisition route; The minimum bounding sphere drawing conditions include: The direction angles between the two facing directions of the surface of the area of ​​the detection area in the minimum enclosing sphere all fall within a preset angle range; The minimum distances between any two of the surface areas are less than or equal to a preset first distance threshold; On each of the regional surfaces, there exists a target straight line parallel to the same diameter of the minimum enclosing sphere; both ends of the target straight line fall on the boundary line of the regional surface, and the minimum distance between the surface center position of the regional surface and the target straight line is less than or equal to a preset second distance threshold; The volume of the minimum enclosing sphere is less than or equal to a preset volume threshold; The first differences between the detection priority values ​​of the detection areas in the minimum enclosing sphere are all less than or equal to a preset first difference threshold; The detection order value determination rules include: Determine a maximum priority value and a minimum priority value among the detection priority values ​​of the detection area in the minimum enclosing sphere; When the second difference between the maximum priority value and the minimum priority value is greater than or equal to a preset second difference threshold, the maximum priority value is used as the detection order value; otherwise, the average value of the detection priority values ​​of the detection area in the minimum enclosing sphere is used as the detection order value.

2. A machine vision-based bridge crack detection system as claimed in claim 1, characterized in that: The image acquisition route planning module obtains the detection priority values ​​of multiple detection areas in the crack detection map and performs the following operations: Determining the region information of the detection region from the crack detection map; Based on a preset first feature extraction template, feature extraction is performed on the region information to obtain an information feature set; Obtaining a preset detection priority value determination library; The detection priority value determination library includes: multiple sets of one-to-one corresponding standard information feature sets and standard values; Determining whether the information feature set includes any of the standard information feature sets; When the answer is yes, the standard value corresponding to the standard information feature set included in the information feature set is used as the target value; Each of the target values ​​is cumulatively calculated to obtain the detection priority value of the detection area.

3. A bridge crack detection system based on machine vision as claimed in claim 1, characterized in that: The image acquisition route planning module obtains the shooting position and shooting angle corresponding to the minimum bounding sphere and performs the following operations: Constructing a first direction vector based on the surface center position of the area surface of the detection area in the minimum enclosing sphere and the surface facing direction; Setting a preset virtual drone corresponding to the drone next to the minimum enclosing sphere; Continuously obtaining the flight position and shooting direction of the virtual drone; constructing a second direction vector based on the flight position and the shooting direction; Performing flight adjustment on the virtual drone; when performing flight adjustment on the virtual drone, the vector angle between any one of the first direction vectors and the second direction vector falls within a preset vector angle range; Until the maximum shooting range of the virtual drone just includes the surface of the entire detection area in the minimum enclosing sphere, the current flight position of the virtual drone is used as the shooting position corresponding to the minimum enclosing sphere, and the current shooting direction of the virtual drone is used as the shooting angle corresponding to the minimum enclosing sphere.

4. The machine vision-based bridge crack detection system according to claim 1, characterized in that: The crack detection module performs crack detection based on machine vision and the collected images, and performs the following operations: Based on a preset second feature extraction template, feature extraction is performed on the collected image to obtain a plurality of image features; Matching the image features with standard image features in a preset standard image feature library; When the match is satisfied, it is determined that a bridge crack appears on the corresponding acquired image.

5. A bridge crack detection method based on machine vision, characterized in that: include: Step S1: obtaining a preset crack detection map corresponding to a target bridge; Step S2: planning an image acquisition route based on the crack detection map; Step S3: Based on the image acquisition route, controlling the UAV to acquire images of the target bridge to obtain acquired images; Step S4: Based on machine vision, crack detection is performed according to the collected image, and the detection result is output; The step S2: planning an image acquisition route based on the crack detection map, comprises: Obtaining detection priority values ​​of multiple detection areas in the crack detection map; Based on a minimum bounding sphere drawing condition, drawing a minimum bounding sphere including at least two of the detection areas in the crack detection map; Obtaining a shooting position and a shooting angle corresponding to the minimum enclosing sphere; Based on a detection order value determination rule, determining a detection order value according to the detection priority value of the detection area in the minimum enclosing sphere; The corresponding shooting positions are sorted from large to small according to the detection order values ​​to obtain a position sequence; Planning a flight route between each of the shooting positions in the position sequence in the crack detection map; Connecting the flight routes end to end and marking the shooting angles at the corresponding shooting positions on the flight routes to obtain the image acquisition route; The minimum bounding sphere drawing conditions include: The direction angles between the two facing directions of the surface of the area of ​​the detection area in the minimum enclosing sphere all fall within a preset angle range; The minimum distances between any two of the surface areas are less than or equal to a preset first distance threshold; On each of the regional surfaces, there exists a target straight line parallel to the same diameter of the minimum enclosing sphere; both ends of the target straight line fall on the boundary line of the regional surface, and the minimum distance between the surface center position of the regional surface and the target straight line is less than or equal to a preset second distance threshold; The volume of the minimum enclosing sphere is less than or equal to a preset volume threshold; The first differences between the detection priority values ​​of the detection areas in the minimum enclosing sphere are all less than or equal to a preset first difference threshold; The detection order value determination rules include: Determine a maximum priority value and a minimum priority value among the detection priority values ​​of the detection area in the minimum enclosing sphere; When the second difference between the maximum priority value and the minimum priority value is greater than or equal to a preset second difference threshold, the maximum priority value is used as the detection order value; otherwise, the average value of the detection priority values ​​of the detection area in the minimum enclosing sphere is used as the detection order value.

6. A bridge crack detection method based on machine vision as claimed in claim 5, characterized in that: The obtaining of detection priority values ​​of multiple detection areas in the crack detection map comprises: Determining the region information of the detection region from the crack detection map; Based on a preset first feature extraction template, feature extraction is performed on the region information to obtain an information feature set; Obtaining a preset detection priority value determination library; the detection priority value determination library includes: multiple sets of one-to-one corresponding standard information feature sets and standard values; Determining whether the information feature set includes any of the standard information feature sets; When the answer is yes, the standard value corresponding to the standard information feature set included in the information feature set is used as the target value; Each of the target values ​​is cumulatively calculated to obtain the detection priority value of the detection area.

7. The bridge crack detection method based on machine vision as claimed in claim 5, characterized in that: The obtaining of the shooting position and shooting angle corresponding to the minimum enclosing sphere includes: Constructing a first direction vector based on the surface center position of the area surface of the detection area in the minimum enclosing sphere and the surface facing direction; Setting a preset virtual drone corresponding to the drone next to the minimum enclosing sphere; Continuously obtaining the flight position and shooting direction of the virtual drone; constructing a second direction vector based on the flight position and the shooting direction; Performing flight adjustment on the virtual drone; when performing flight adjustment on the virtual drone, the vector angle between any one of the first direction vectors and the second direction vector falls within a preset vector angle range; Until the maximum shooting range of the virtual drone just includes the surface of the entire detection area in the minimum enclosing sphere, the current flight position of the virtual drone is used as the shooting position corresponding to the minimum enclosing sphere, and the current shooting direction of the virtual drone is used as the shooting angle corresponding to the minimum enclosing sphere.

8. The bridge crack detection method based on machine vision as claimed in claim 5, characterized in that: The step S4: performing crack detection based on the collected image based on machine vision, comprises: Based on a preset second feature extraction template, feature extraction is performed on the collected image to obtain a plurality of image features; Matching the image features with standard image features in a preset standard image feature library; When the match is satisfied, it is determined that a bridge crack appears on the corresponding acquired image.

Images