Automatic analysis method of sidewalk damage range based on laser point cloud

Abstract

The present invention provides a method for automatically analyzing the scope of sidewalk damage based on laser point cloud, comprising collecting raw data through a scanner to generate point cloud data of a sidewalk area; filtering the collected point cloud data to remove noise and abnormal values, performing motion distortion compensation and point cloud framing on the point cloud, and generating a geographic reference image using elevation information, intensity information and point cloud density information of the point cloud data; wherein the geographic reference image includes an elevation image, an intensity image and a point density image; preliminarily determining the sidewalk edge and the damage area based on the geographic reference image; based on the preliminarily determined sidewalk edge and the damage area, using an improved active contour model algorithm to accurately analyze the damage scope. The present invention can more efficiently and accurately identify and analyze the damage area from the laser point cloud data, thereby improving the automation level and maintenance efficiency of sidewalk damage detection.

Description

Technical Field

[0001] The invention relates to the technical field of computer vision and pattern recognition, and in particular to an automatic analysis method of a sidewalk disease range based on laser point cloud. Background Art

[0002] Under the influence of vehicle load, temperature, ultraviolet rays, etc., the road pavement structure will inevitably suffer damage as the operating time increases, such as cracks, subsidence, and interlayer voids. Quickly, effectively, and accurately finding the location and type of pavement structure defects, determining their severity, and predicting the development of the disease are of great significance for scientifically and rationally determining the maintenance opportunity and formulating maintenance plans.

[0003] At present, sidewalk defects are detected manually by street or transportation departments, which requires a lot of manpower and has low efficiency. In addition, the existing sidewalk defect detection method based on laser point cloud has limitations in accuracy and efficiency when dealing with complex environments and large amounts of data, especially in the automatic analysis of the scope of the defect. Summary of the invention

[0004] 1. Technical issues to be resolved

[0005] In view of the deficiencies in the prior art, the present invention provides a method for automatically analyzing the scope of sidewalk damage based on laser point cloud to solve the problems raised in the above background technology.

[0006] (II) Technical solution

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: a method for automatically analyzing the scope of sidewalk damage based on laser point cloud, the method comprising:

[0008] The original data is collected by the scanner to generate point cloud data of the sidewalk area;

[0009] Filter the collected point cloud data to remove noise and outliers, perform motion distortion compensation and point cloud framing on the point cloud, and generate georeferenced images using the elevation information, intensity information, and point cloud density information of the point cloud data;

[0010] Among them, georeferenced images include elevation images, intensity images, and point density images;

[0011] Preliminary identification of sidewalk edges and damaged areas based on geo-referenced images;

[0012] Based on the initially determined sidewalk edges and damaged areas, an improved active contour model algorithm is used to accurately analyze the damage range.

[0013] The damaged area is output in the form of a three-dimensional point cloud and the polygonal boundary of each damaged area is generated in two-dimensional space. At the same time, the judgment data of the sidewalk is calculated, and the maintenance decision is output based on the judgment data.

[0014] As a preferred embodiment of this embodiment, the process of determining the sidewalk edge and the damaged area based on the geographic reference image includes:

[0015] Preliminary segmentation of ground and edge: Calculate the covariance matrix for each point in the point cloud, analyze the normal vector through PCA, analyze the planarity based on the eigenvalue, extract the ground area, and use the normal vector to determine whether the area is a plane or an edge. Mark the area with low height and strong planarity as sidewalk or ground, and further process the edge or height mutation area;

[0016] Accurate detection of edges and damaged areas: Use star search to detect complex edge shapes, emit rays and detect the intersection of rays and edges, combine X-zero and Z-zero methods to perform sensitive detection of local height or level changes and identify damaged areas.

[0017] As a preferred embodiment of the present invention, the process of calculating the covariance matrix for each point of the point cloud, analyzing the normal vector by PCA, analyzing the planarity according to the eigenvalue, extracting the ground area, and using the normal vector to determine whether the area is a plane or an edge includes: for each point and its neighborhood points, calculate the centroid μ of these neighborhood points:

[0018]

[0019] Where N is the number of neighborhood points;

[0020] Based on the centroid, calculate the covariance matrix C of the point cloud neighborhood:

[0021]

[0022] Perform eigenvalue decomposition on the covariance matrix C and obtain three eigenvalues ≥ ≥ and its corresponding eigenvector , , ;

[0023] Eigenvector The corresponding eigenvalue Indicates the direction of minimum change in the area, corresponding to the normal vector of the ground;

[0024] if < ,and ≈ , then the region can be regarded as a plane;

[0025] when > , indicating that the point does not belong to the plane, where Indicates the threshold value.

[0026] As a preferred embodiment of this embodiment, the process of accurately analyzing the damage range using the improved active contour model algorithm based on the detected damage area includes:

[0027] Based on the preliminary detection results of the diseased area, a contour is initialized around the diseased area;

[0028] By iteratively optimizing the total energy, the position of the contour is gradually adjusted to minimize its energy;

[0029] Among them, the contour is attracted to the edge of the disease by the external energy, while the internal energy maintains the smoothness of the contour, and the regional energy helps to identify the internal complex structure;

[0030] When the total energy function converges to the local minimum or the contour variation amplitude tends to be stable, the contour reaches the precise boundary of the diseased area and the final contour is obtained;

[0031] The final contour is used as the boundary of the diseased area to parse out the information of the diseased area;

[0032] The final contour can be converted into a region mask in a 3D point cloud or a 2D image.

[0033] As a preference of this embodiment, the total energy function of the active contour model is specifically:

[0034]

[0035] in, represents the active contour, represents the total energy, Represents the internal energy, which is used to control the smoothness and continuity of the curve. Represents external energy, which attracts the contour to the edge of the image based on the image gradient information. Represents regional energy, based on the internal characteristics of the diseased area, and helps the contour fit the diseased area more accurately through regional statistical information.

[0036] α, β, and γ represent three weight parameters, which respectively control the contribution of internal energy, external energy, and regional energy to the total energy.

[0037] As a preferred embodiment of this embodiment, the internal energy is used to ensure the smoothness and continuity of the curve, and its calculation formula is:

[0038]

[0039] Represents active contour It follows the parameters The changing curve, The first derivative of the curve, The second derivative of a curve.

[0040] As a preferred embodiment of this invention, the external energy is used to pull the active contour to the edge of the image, and the calculation formula is:

[0041]

[0042] Indicates that image I is at the boundary The gradient at .

[0043] As a preferred embodiment of this invention, the regional energy is used to enhance the model to analyze the internal characteristics of the diseased area more accurately, and the calculation formula is:

[0044]

[0045] Represents the pixel value of the image at point (x, y), Represents the average intensity or other characteristics of the pixels inside the diseased area.

[0046] As a preferred embodiment of the present invention, the process of calculating the determination data of the sidewalk includes: according to the output two-dimensional polygon description, the area of ​​the diseased area can be calculated. , the calculation formula of disease area is:

[0047]

[0048] Where dA is the small area unit within the polygon; the sidewalk damage density D is calculated based on the area of ​​the diseased area, and the calculation formula is:

[0049]

[0050] in, Represents the total area of ​​the diseased area detected, Represents the total area of ​​the sidewalk; calculates the sidewalk integrity rate based on the area of ​​the diseased area The calculation formula is:

[0051] .

[0052] As a preferred embodiment of this invention, the process of outputting a maintenance decision based on the determination data includes: determining a sidewalk condition index based on the determination data:

[0053]

[0054] in:

[0055] Indicates The individual deduction points for the type of damage; represents the sidewalk condition index; where, The value range is [0,100], 100 means intact, and the lower the value, the more serious the damage.

[0056] (III) Beneficial effects

[0057] The present invention provides a method for automatically parsing the scope of sidewalk damage based on laser point cloud, which has the following beneficial effects: the present invention provides overall segmentation of a large range of ground and sidewalks through covariance matrix and PCA, extracts ground and non-ground areas through geometric features, and further detects local edges and damage areas in combination with star search, X-zero and Z-zero methods, so as to achieve accurate detection of sidewalk edges and damage areas. The entire process can effectively identify sidewalk edge and surface damage, and the damage area after active contour analysis can generate a three-dimensional point cloud to mark the volume and contour of the damage; by marking the contour of the disease boundary on a two-dimensional image, a mask image of the disease is generated, so that the disease area is intuitively visible in the image; the geometric information of the contour can be used to calculate the area and perimeter of the disease, and then used for quantitative analysis of the disease, so as to achieve more efficient and accurate identification and parsing of the damage area from the laser point cloud data, and improve the automation level and maintenance efficiency of sidewalk disease detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 This is a flow chart of the method for automatically analyzing the sidewalk disease range based on laser point cloud of the present invention;

[0059] Figure 2 This is a schematic diagram of the scanning point space measurement principle when collecting raw data of the present invention. DETAILED DESCRIPTION

[0060] The embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0061] The disclosure below provides many different embodiments or examples to realize different structures of the present invention. In order to simplify the disclosure of the present invention, the parts and settings of specific examples are described below. Of course, they are only examples, and the purpose is not to limit the present invention. In addition, the present invention can repeat reference numbers and / or reference letters in different examples, and this repetition is for the purpose of simplicity and clarity, which itself does not indicate the relationship between the various embodiments and / or settings discussed. In addition, the present invention provides various specific examples of processes and materials, but those of ordinary skill in the art can be aware of the application of other processes and / or the use of other materials.

[0062] like Figure 1 As shown, an embodiment of the present invention provides a method for automatically analyzing the sidewalk disease range based on laser point cloud, the method comprising:

[0063] Step 1: Collect raw data through a scanner to generate point cloud data of the sidewalk area;

[0064] Among them, the handheld scanner is based on the SLAM algorithm, i.e., simultaneous positioning and mapping, and corrects point cloud data. It carries its own sensors (encoders, IMU, etc.). Even in an unknown underground space, it can combine external sensors for self-positioning, real-time correction and evaluation of the scanning probe's posture, reduce the impact of the surveyor's travel speed on the measurement quality, and use external sensors to obtain underground space information in a spiral incremental manner to build an unknown environment map. The principle of laser probe ranging is to automatically emit an infrared beam, contact the surface of the object, and the odometer records the distance between the laser emission point and the object. The laser beam is immediately reflected back to the scanner, and finally fits with the horizontal and vertical angles to obtain the three-dimensional coordinates of each point, and quickly construct line, surface, body and other map data, which is similar to the ranging principle of the total station. Let the observation distance (measurement point-target point) be S, C be the speed, and t be the time variable, then S=1 / 2×C×t2. The coordinates of the laser scanning point can be calculated according to the following formula:

[0065]

[0066] In the above formula, φ and θ are the observed values ​​of scanning angles in the horizontal and vertical directions respectively. The X-axis is in the horizontal scanning, the Z-axis is perpendicular to the horizontal scanning plane, and the Y-axis is perpendicular to the X-axis and in the vertical scanning plane. The principle of scanning point space measurement is as follows: Figure 2 As shown;

[0067] In addition, during scanning and measurement, people have limited physical strength and cannot always maintain a posture in a straight line without deviation. At this time, the inertial measurement unit (IMU) inside the system starts to work, correcting the posture and direction of travel, and using Newton's laws of motion to measure the angular velocity and acceleration information of the human body relative to the inertial space, plus time integration operations, to obtain the person's instantaneous speed and position information. The scanner and the inertial measurement unit are not in the same coordinate system when measuring the target object. The data of the two need to be fused and unified in the same coordinate system to obtain the spatial position. In addition, the coordinate centers of each part are not absolutely unified in space, and there is a certain difference between them ( ).

[0068] Let O be the origin of the coordinate system, transform the scanner measurement coordinate system to the inertial measurement unit coordinate system, and add the difference =( , , ). Use the matrix formula to calculate the coordinate position of the scanning point in the new coordinate system ( , , ), get the unified spatial coordinate information, the conversion formula is shown as follows:

[0069]

[0070] In the formula,

[0071]

[0072]

[0073]

[0074] In the above matrix formula, α, β, and γ are the relative conversion angles of the X-axis, Y-axis, and Z-axis when the laser scanner coordinate system is converted to the inertial measurement unit coordinate system. The coordinate conversion of other components is obtained in the same way, that is, they are all unified to the coordinate system of the inertial measurement unit to generate the initial coordinate information in the scanner.

[0075] Step 2: Filter the collected point cloud data to remove noise and outliers, perform motion distortion compensation and point cloud framing on the point cloud, and generate a georeferenced image using the elevation information, intensity information, and point cloud density information of the point cloud data;

[0076] Among them, georeferenced images include elevation images, intensity images, and point density images;

[0077] The data processing process includes: filtering the point cloud to remove noise and outliers; downsampling the data using the voxel grid downsampling method to reduce computational complexity; and performing motion distortion compensation and point cloud framing.

[0078] Step 3: Preliminary determination of sidewalk edges and damaged areas based on georeferenced images;

[0079] Step 4: Based on the initially determined sidewalk edge and damaged area, the improved active contour model algorithm is used to accurately analyze the damage range;

[0080] Step 5: Output the damaged area in the form of a three-dimensional point cloud and generate a polygonal boundary for each damaged area in two-dimensional space. At the same time, calculate the judgment data of the sidewalk and output the maintenance decision based on the judgment data.

[0081] Furthermore, the process of determining the sidewalk edge and the damaged area based on the geo-referenced image includes:

[0082] Preliminary segmentation of ground and edge: Calculate the covariance matrix for each point in the point cloud, analyze the normal vector through PCA, analyze the planarity based on the eigenvalue, extract the ground area, and use the normal vector to determine whether the area is a plane or an edge. Mark the area with low height and strong planarity as sidewalk or ground, and further process the edge or height mutation area;

[0083] Accurate detection of edges and damaged areas: Use star search to detect complex edge shapes, emit rays and detect the intersection of rays and edges, combine X-zero and Z-zero methods to perform sensitive detection of local height or level changes and identify damaged areas.

[0084] Among them, star search is used for edge detection, which is particularly suitable for edges of complex shapes. The basic principle is to start from the center point Shoot rays in different directions (usually radially) and find the closest point in each direction :

[0085]

[0086] When a ray intersects the boundary of the target, the point detected By emitting rays from multiple directions, the point set of the entire boundary can be obtained.

[0087] The X-zero and Z-zero methods are used to detect sudden changes along the x-axis and z-axis:

[0088] The X-zero method is used to detect depth changes along the horizontal direction (x-axis).

[0089] The Z-zero method is used to detect sudden changes in height along the z-axis and is often used to detect surface defects such as cracks, sags, and bulges.

[0090] For two adjacent points and , their coordinate difference is:

[0091]

[0092] When |Δx|> or |Δz|> ( and is the set threshold), indicating that a local mutation was detected, which may be an edge or diseased area.

[0093] It is understandable that by setting an appropriate threshold and To mark points of local height and level mutations as possible disease areas.

[0094] Furthermore, the process of calculating the covariance matrix for each point of the point cloud, analyzing the normal vector by PCA, analyzing the planarity according to the eigenvalue, extracting the ground area, and using the normal vector to determine whether the area is a plane or an edge includes:

[0095] For each point and its neighborhood points, calculate the centroid μ of these neighborhood points:

[0096]

[0097] Where N is the number of neighborhood points;

[0098] Based on the centroid, calculate the covariance matrix C of the point cloud neighborhood:

[0099]

[0100] Perform eigenvalue decomposition on the covariance matrix C and obtain three eigenvalues ≥ ≥ and its corresponding eigenvector , , ;

[0101] Eigenvector The corresponding eigenvalue Indicates the direction of minimum change in the area, corresponding to the normal vector of the ground;

[0102] if < ,and ≈ , then the region can be regarded as a plane;

[0103] when > , indicating that the point does not belong to the plane, where Indicates the threshold value.

[0104] By combining this method, the covariance matrix and PCA provide an overall segmentation of the large-scale ground and sidewalk, while the star search, X-zero and Z-zero methods further detect local edges and diseased areas. The entire process can effectively identify sidewalk edges and surface diseases, and intuitively display the detection results in the form of images or point clouds.

[0105] Furthermore, the process of accurately analyzing the scope of the disease using the improved active contour model algorithm based on the detected disease area includes:

[0106] Based on the preliminary detection results of the diseased area, a contour is initialized around the diseased area;

[0107] By iteratively optimizing the total energy, the position of the contour is gradually adjusted to minimize its energy;

[0108] Among them, the contour is attracted to the edge of the disease by the external energy, while the internal energy maintains the smoothness of the contour, and the regional energy helps to identify the internal complex structure;

[0109] When the total energy function converges to the local minimum or the contour variation amplitude tends to be stable, the contour reaches the precise boundary of the diseased area and the final contour is obtained;

[0110] The final contour is used as the boundary of the diseased area to parse out the information of the diseased area;

[0111] The final contour can be converted into a region mask in a 3D point cloud or a 2D image.

[0112] Furthermore, the total energy function of the active contour model is specifically:

[0113]

[0114] in, represents the active contour, represents the total energy, Represents the internal energy, which is used to control the smoothness and continuity of the curve. Represents external energy, which attracts the contour to the edge of the image based on the image gradient information. Represents regional energy, based on the internal characteristics of the diseased area, and helps the contour fit the diseased area more accurately through regional statistical information.

[0115] α, β, and γ represent three weight parameters, which respectively control the contribution of internal energy, external energy, and regional energy to the total energy.

[0116] As a preferred embodiment of this embodiment, the internal energy is used to ensure the smoothness and continuity of the curve, and its calculation formula is:

[0117]

[0118] Represents active contour It follows the parameters The changing curve, The first derivative of the curve, The second derivative of a curve.

[0119] Furthermore, the external energy is used to pull the active contour to the edge of the image, and its calculation formula is:

[0120]

[0121] Indicates that image I is at the boundary The gradient at .

[0122] As a preferred embodiment of this invention, the regional energy is used to enhance the model to analyze the internal characteristics of the diseased area more accurately, and the calculation formula is:

[0123]

[0124] Represents the pixel value of the image at point (x, y), Represents the average intensity or other characteristics of the pixels inside the diseased area.

[0125] It can be understood that the diseased area after active contour analysis can generate a three-dimensional point cloud to mark the volume and contour of the disease; by marking the contour of the disease boundary on the two-dimensional image, a mask image of the disease is generated, making the diseased area intuitively visible in the image; the geometric information of the contour can be used to calculate the area and perimeter of the disease, and then used for quantitative analysis of the disease.

[0126] Among them, 3D voxel point cloud: output 3D point cloud of the damaged area, including the volume, depth, location and other information of the damage. These data can be imported into the road maintenance management system for subsequent analysis.

[0127] Two-dimensional polygon description: The two-dimensional polygon results of the diseased area can be used for mapping, report generation and further geographic information analysis.

[0128] Furthermore, the process of calculating the determination data of the sidewalk includes:

[0129] According to the output two-dimensional polygon description, the area of ​​the diseased area can be calculated , the calculation formula of disease area is:

[0130]

[0131] Where dA is the small area unit within the polygon; the sidewalk damage density D is calculated based on the area of ​​the diseased area, and the calculation formula is:

[0132]

[0133] in, represents the total area of ​​the diseased area detected, Represents the total area of ​​the sidewalk; calculates the sidewalk integrity rate based on the area of ​​the diseased area The calculation formula is:

[0134] .

[0135] Among them, according to the P value, the maintenance condition of the sidewalk can be divided into the following four levels:

[0136]

[0137] Further, the process of outputting a maintenance decision based on the determination data includes: determining a sidewalk condition index based on the determination data:

[0138]

[0139] in:

[0140] Indicates The individual deduction points for the type of damage; represents the sidewalk condition index; where, The value range is [0,100], 100 means intact, and the lower the value, the more serious the damage.

[0141]

[0142] The above table is a table of deduction points for individual items of sidewalk pavement damage (unit: points)

[0143] It should be noted that according to the FCI value, the damage condition of the sidewalk can be divided into the following four levels:

[0144]

[0145] Based on the calculated FCI value and the specific damage level of the pavement, relevant repair decisions can be made:

[0146] FCI ≥ 80: The sidewalk is in good condition, with minor damage and no immediate repairs required. Regular inspection and maintenance and minor repairs are acceptable.

[0147] 65 ≤ FCI<80: The sidewalk is slightly damaged and requires preventive maintenance, minor repairs or medium repairs, such as simple patching, cleaning and protective treatment.

[0148] 50 ≤ FCI<65: Moderate damage, requiring medium or partial repairs, such as repairing cracks, looseness or deformation to ensure pedestrian safety and the service life of the sidewalk.

[0149] FCI<50: Severely damaged, in urgent need of major repairs or renovation and expansion projects, comprehensive renovations or emergency repairs, especially those involving large areas of surface damage or structural problems, must be given priority.

[0150] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for automatically analyzing the scope of sidewalk damage based on laser point cloud, characterized in that: The method comprises: The original data is collected by the scanner to generate point cloud data of the sidewalk area; Filter the collected point cloud data to remove noise and outliers, perform motion distortion compensation and point cloud framing on the point cloud, and generate georeferenced images using the elevation information, intensity information, and point cloud density information of the point cloud data; Among them, georeferenced images include elevation images, intensity images, and point density images; Preliminary identification of sidewalk edges and damaged areas based on geo-referenced images; Based on the initially determined sidewalk edges and damaged areas, an improved active contour model algorithm is used to accurately analyze the damage range. Output the damaged area in the form of a 3D point cloud and generate the polygonal boundary of each damaged area in 2D space. Calculate the judgment data of the sidewalk and output maintenance decisions based on the judgment data. Based on the detected diseased area, the process of using the improved active contour model algorithm to accurately analyze the disease range includes: Based on the preliminary detection results of the diseased area, a contour is initialized around the diseased area; By iteratively optimizing the total energy, the position of the contour is gradually adjusted to minimize its energy; Among them, the contour is attracted to the edge of the disease by the external energy, while the internal energy maintains the smoothness of the contour, and the regional energy helps to identify the internal complex structure; When the total energy function converges to the local minimum or the contour variation amplitude tends to be stable, the contour reaches the precise boundary of the diseased area and the final contour is obtained; The final contour is used as the boundary of the diseased area to parse out the information of the diseased area; The final contour can be converted into a region mask in a 3D point cloud or a 2D image.

2. The method for automatically analyzing the sidewalk damage range based on laser point cloud according to claim 1 is characterized in that: The process of determining the sidewalk edge and the damaged area based on the geo-referenced image includes: Preliminary segmentation of ground and edge: Calculate the covariance matrix for each point in the point cloud, analyze the normal vector through PCA, analyze the planarity based on the eigenvalue, extract the ground area, and use the normal vector to determine whether the area is a plane or an edge. Mark the area with low height and strong planarity as sidewalk or ground, and further process the edge or height mutation area; Accurate detection of edges and damaged areas: Use star search to detect complex edge shapes, emit rays and detect the intersection of rays and edges, combine X-zero and Z-zero methods to perform sensitive detection of local height or level changes and identify damaged areas.

3. The method for automatically analyzing the sidewalk damage range based on laser point cloud according to claim 2 is characterized in that: The process of calculating the covariance matrix for each point of the point cloud, analyzing the normal vector by PCA, analyzing the planarity according to the eigenvalue, extracting the ground area, and using the normal vector to determine whether the area is a plane or an edge includes: For each point and its neighborhood points, calculate the centroid μ of these neighborhood points: Where N is the number of neighborhood points; Based on the centroid, calculate the covariance matrix C of the point cloud neighborhood: Perform eigenvalue decomposition on the covariance matrix C and obtain three eigenvalues ≥ ≥ and its corresponding eigenvector , , ; Eigenvector The corresponding eigenvalue Indicates the direction of minimum change in the area, corresponding to the normal vector of the ground; if < ,and ≈ , then the region can be regarded as a plane; when > , it means that the point does not belong to the plane, where represents the threshold.

4. The method for automatically analyzing the sidewalk damage range based on laser point cloud according to claim 1 is characterized in that: The total energy function of the active contour model is specifically: in, represents the active contour, represents the total energy, Represents the internal energy, which is used to control the smoothness and continuity of the curve. Represents external energy, which attracts the contour to the edge of the image based on the image gradient information. Represents regional energy, based on the internal characteristics of the diseased area, and helps the contour fit the diseased area more accurately through regional statistical information. ɑ, β, and γ represent three weight parameters, which respectively control the contribution of internal energy, external energy, and regional energy to the total energy.

5. The method for automatically analyzing the sidewalk damage range based on laser point cloud according to claim 4 is characterized by: The internal energy is used to ensure the smoothness and continuity of the curve, and its calculation formula is: Represents active contour It follows the parameters The changing curve, : The first derivative of the curve, : The second derivative of the curve.

6. The method for automatically analyzing the sidewalk damage range based on laser point cloud according to claim 5 is characterized in that: The external energy is used to pull the active contour to the edge in the image, and its calculation formula is: Representing images At the border The gradient at .

7. The method for automatically analyzing the sidewalk damage range based on laser point cloud according to claim 5 is characterized in that: The regional energy is used to enhance the model to analyze the internal characteristics of the diseased area more accurately, and its calculation formula is: Indicates that the image is at point The pixel value at Represents the average intensity or other characteristics of the pixels inside the diseased area.

8. The method for automatically analyzing the sidewalk damage range based on laser point cloud according to claim 1 is characterized in that: The process of calculating the determination data of the sidewalk includes: According to the output two-dimensional polygon description, the area of ​​the diseased area can be calculated , the calculation formula of disease area is: in It is a tiny unit of area within a polygon; Calculate sidewalk damage density based on the area of ​​the diseased area , the calculation formula is: in, represents the total area of ​​the diseased area detected, represents the total area of ​​sidewalks; Calculate the pavement integrity rate based on the area of ​​the diseased area , the calculation formula is: .

9. The method for automatically analyzing the sidewalk damage range based on laser point cloud according to claim 8 is characterized in that: The process of outputting a maintenance decision based on the determination data includes: determining a sidewalk condition index based on the determination data: in: Indicates The individual deduction points for the type of damage; represents the sidewalk condition index; in, The value range is [0,100], 100 means intact, and the lower the value, the more serious the damage.

Images