Off-line road crack detection method and system based on crack self-growth

Abstract

The invention discloses an off-line road crack detection method and system based on crack self-growth, and the method comprises the steps: obtaining a road image, segmenting the road image into a plurality of original regions with closed boundaries, and calculating the disturbance quantity of texture changes in each original region; judging whether each original area is an activated area or not according to the disturbance quantity, identifying a crack starting point in each activated area, and deleting other original areas; and starting from the crack starting point in the activated area, growing a crack track according to the disturbance direction of the disturbance quantity in the activated area until the growth of the crack track is completed, thereby completing the road crack detection. According to the method, the road image is segmented into the original areas with the closed boundaries, the disturbance quantity of texture change of each area is calculated, then the crack growth track is dynamically simulated according to the disturbance direction, autonomous extension detection from the starting point to the terminal is achieved, and the detection precision is improved to 95% or above.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of road crack detection, and more particularly relates to an offline road crack detection method and system based on crack self-growth. BACKGROUND

[0002] Currently, road crack detection, as a key link in infrastructure maintenance and management, has become an important research direction in intelligent transportation and smart city construction. Traditional road crack detection methods mainly rely on manual inspection, and detection personnel observe the road surface conditions by naked eye and manually record the crack type, length and position. Although this method is intuitive, it is inefficient, labor-intensive, and easily affected by subjective factors, with unstable detection accuracy and high missed detection rate, which is difficult to meet the growing demand for road maintenance. Especially in large-scale road networks, manual detection cannot meet the demand for high-frequency and high-accuracy crack identification and evaluation.

[0003] Current mainstream detection technologies can be divided into two categories: one is manual inspection, which relies on detection personnel to visually identify cracks and manually record types and positions, and is significantly affected by lighting conditions and subjective experience; the other is based on image recognition technology, such as using vehicle-mounted / unmanned aerial vehicle-mounted high-definition camera equipment to collect road surface images, and using U-Net, Mask R-CNN, etc. algorithm for crack semantic segmentation, commercial systems such as RoadBotics, CrackMap typically have an identification accuracy of only 82%-86%. With the development of image processing and artificial intelligence technology, automatic crack detection based on images has gradually become a research hotspot.

[0004] This method usually uses high-definition camera equipment (such as industrial cameras, vehicle-mounted cameras or unmanned aerial vehicle-mounted cameras) to collect road surface images, extracts crack features through image processing algorithms, and then identifies and classifies cracks. However, the existing technology still has the following technical problems: (1) low recognition accuracy and inability to make cracks grow autonomously to complete detection; (2) recognition accuracy bottleneck, image algorithms have low sensitivity to fine cracks, with a missed detection rate of up to 18%; noise interference leads to a false detection rate of more than 15%; (3) lack of dynamic prediction, existing technology only realizes static crack identification, and cannot simulate crack evolution trends (such as predicting the expansion direction and rate), resulting in a lack of basis for preventive maintenance decisions; SUMMARY

[0005] To solve the above technical problems, the present application provides an offline road crack detection method and system based on crack self-growth, which divides the road image into boundary-closed original regions, calculates the disturbance of texture change in each region, screens out the regions with disturbance exceeding the threshold as active regions and locates the crack starting point; then, the crack growth trajectory is dynamically simulated according to the disturbance direction to realize autonomous extension detection from the starting point to the terminal. This mechanism breaks through the limitation of traditional static recognition, introduces crack evolution dynamics into the detection process, solves the industry pain point of high missing detection rate of complex pavement, provides data basis for predicting crack expansion trend, improves the detection accuracy to more than 95%, can accurately complete the self-growth of road crack trajectory, and can grow together with multiple branches, thereby improving the road crack detection efficiency.

[0006] To achieve the above purpose, according to the first aspect of the present application, a kind of offline road crack detection method based on crack self-growth is provided, comprising:

[0007] obtain road image, the road image is divided into multiple boundary-closed original regions, and the disturbance of texture change in each original region is calculated;

[0008] determine whether each original region is an active region according to the disturbance, and identify the crack starting point in each active region, wherein the remaining original regions are deleted;

[0009] from the crack starting point in the active region, grow crack trajectory according to the disturbance direction of the disturbance in the active region until crack trajectory growth is completed, thereby completing road crack detection.

[0010] Further, the road image is divided into multiple boundary-closed original regions, comprising:

[0011] convert the road image into a gray image;

[0012] extract the structure direction tensor matrix of each point in the gray image, obtain the eigenvalue of the structure direction tensor matrix of each point, and calculate the direction vector of each point through the eigenvalue;

[0013] calculate the main direction direction vector of each point according to the structure direction tensor matrix and the direction vector, and convert the main direction direction vector into a main direction direction angle;

[0014] calculate the local direction change rate of the main direction direction angle for describing the mutation degree of main direction, when the local direction change rate is greater than the direction change threshold, the current point is a structure change boundary, traverse all points on the gray image, find all structure change boundary points, and perform connected region merging on non-structure change boundary points, each connected region as a piece of original region.

[0015] Further, judging whether each of the original regions is an active region according to the disturbance amount comprises:

[0016] Calculating a pixel standard deviation of each of the original regions, and deleting the original region corresponding to the pixel standard deviation less than a preset standard deviation threshold value to avoid a false crack in a shadow, pure white or pure black region;

[0017] Obtaining a disturbance vector of the remaining original region, which is composed of a main direction disturbance index, an average edge response intensity and a structure skewness;

[0018] Calculating a total disturbance intensity of the remaining original region according to the disturbance vector, and taking the original region corresponding to the total disturbance intensity greater than a region disturbance intensity threshold value as an active region.

[0019] Further, expanding in the disturbance direction of the disturbance amount of the active region to form a complete crack track comprises:

[0020] Taking an arbitrary point in the active region, calculating a total disturbance intensity of the arbitrary point, and taking the arbitrary point as a crack starting point if the total disturbance intensity of the arbitrary point is greater than a point disturbance intensity threshold value, and growing a crack track from the crack starting point in a crack growth direction of a main direction direction angle of the active region;

[0021] Advancing in the crack growth direction by a fixed step length, judging whether each new position meets a crack track growth condition, continuing to grow if yes, otherwise taking a point of the new position as a crack ending point, and finding a point meeting the crack track growth condition in a neighborhood of the crack starting point with the fixed step length as a radius to continue to grow the crack track, wherein the crack track growth condition is that recalculating the total disturbance intensity of the current point at each new position, and continuing to grow if the total disturbance intensity of the current point exceeds the point disturbance intensity threshold value, a direction change amplitude of the current point is less than a direction change threshold value, and a change amplitude of the total disturbance intensity of the current point is greater than an intensity change threshold value.

[0022] Further, when growing the crack track from the crack starting point, finding all points meeting the crack track growth condition in the neighborhood of the crack starting point with the fixed step length as the radius, and growing the crack track to find all branches of the crack track until each branch finds a crack ending point to complete the crack track growth.

[0023] The application further provides an offline road crack detection system based on crack self-growth, comprising:

[0024] A region division module is configured to obtain a road image, divide the road image into a plurality of original regions with closed boundaries, and calculate a disturbance amount of texture change in each of the original regions.

[0025] a crack starting point identification module, configured to determine whether each of the original regions is an active region according to the disturbance amount, and identify a crack starting point in each of the active regions, wherein the remaining original regions are deleted;

[0026] a crack growth module, configured to grow a crack track from the crack starting point in the active region according to the disturbance direction of the disturbance amount in the active region, until the crack track is grown completely, so as to complete the road crack detection.

[0027] Further, the road image is divided into a plurality of original regions with closed boundaries, including:

[0028] the road image is converted into a gray image;

[0029] a structure direction tensor matrix of each point in the gray image is extracted, an eigenvalue of the structure direction tensor matrix of each point is obtained, and a direction vector of each point is calculated through the eigenvalue;

[0030] a principal direction direction vector of each point is calculated according to the structure direction tensor matrix and the direction vector, and the principal direction direction vector is converted into a principal direction direction angle;

[0031] a local direction change rate of the principal direction direction angle for describing a mutation degree of the principal direction is calculated, when the local direction change rate is greater than a direction change threshold value, the current point is a structure change boundary, all points on the gray image are traversed, all points of the structure change boundary are found out, points of a non-structure change boundary are merged, and each connected region is taken as an original region.

[0032] Further, determining whether each of the original regions is an active region according to the disturbance amount includes:

[0033] a pixel standard deviation of each of the original regions is calculated, and the original region corresponding to the pixel standard deviation less than a preset standard deviation threshold value is deleted, so as to avoid a false crack in a shadow, a pure white or a pure black region;

[0034] a disturbance vector of the remaining original region is obtained, which is composed of a principal direction disturbance index, an average edge response intensity and a structure skewness;

[0035] a total disturbance intensity of the remaining original region is calculated according to the disturbance vector, and the original region corresponding to the total disturbance intensity greater than a region disturbance intensity threshold value is taken as an active region.

[0036] Further, expanding in the disturbance direction of the disturbance amount of the active region to form a complete crack track includes:

[0037] Taking an arbitrary point in the activation region, and calculating the total disturbance intensity of the arbitrary point, if the total disturbance intensity of the arbitrary point is greater than the point disturbance intensity threshold, the arbitrary point is taken as a crack starting point, the main direction direction angle of the activation region is taken as a crack growth direction, and the crack trajectory growth is started from the crack starting point;

[0038] A fixed step length is adopted to advance along the main direction direction angle, it is judged whether each new position meets the crack trajectory growth condition, if yes, the growth is continued, otherwise, the point of the new position is taken as a crack end point, a fixed step length is taken as a radius, and a point meeting the crack trajectory growth condition in the neighborhood of the crack starting point is found to continue the crack trajectory growth, wherein the crack trajectory growth condition is that the total disturbance intensity of the current point is recalculated at each new position, if the total disturbance intensity of the current point exceeds the point disturbance intensity threshold, the direction change amplitude of the current point is less than the direction change threshold, and the change amplitude of the total disturbance intensity of the current point is greater than the intensity change threshold, the growth is continued.

[0039] Further, when the crack trajectory growth is started from the crack starting point, a fixed step length is taken as a radius, all points meeting the crack trajectory growth condition in the neighborhood of the crack starting point are found, and the crack trajectory growth is simultaneously performed to find all branches of the crack trajectory until the crack end point of each branch is found, and the complete crack trajectory growth is performed.

[0040] Overall, compared with the prior art, the above technical scheme conceived by the present application has the following beneficial effects:

[0041] 1. The method of the present application, by segmenting the road image into boundary-closed original regions, calculating the disturbance amount of texture change of each region, screening out the regions with disturbance exceeding the threshold as the activation region and locating the crack starting point, then simulating the crack growth trajectory dynamically according to the disturbance direction, realizing the autonomous extension detection from the starting point to the terminal. This mechanism breaks through the limitations of traditional static recognition, introduces crack evolution dynamics into the detection process, solves the industry pain point of high missing detection rate of complex road surface, at the same time provides data basis for predicting crack expansion trend, the detection accuracy is improved to more than 95%, can accurately complete the self-growth of road crack trajectory, and can grow together with multiple branches, improves the road crack detection efficiency.

[0042] 2. The method of the present application, by calculating the pixel standard deviation of the original region, constructing a disturbance vector composed of the main direction disturbance index, the average edge response intensity and the structure skewness, and fusing the three to calculate the total disturbance intensity, only when the total intensity exceeds the dynamic threshold, it is determined as the activation region, so that the activation region recognition accuracy reaches 97.2%, laying a high-purity starting point for crack growth.

[0043] 3. The method of the present application, starting growth from any point in the activation area that meets the point perturbation intensity threshold, advancing by a step size based on the principal direction angle, if the check fails, searching for a new growth point in the neighborhood with the step size as the radius; synchronously starting multi-branch parallel growth, until all branches terminate at the crack end point that does not meet the condition. This mechanism breaks through the limitation of traditional single-path growth, completely captures the bifurcation crack topology, reduces the missed detection rate to <5%, and increases the processing speed by 3 times. BRIEF DESCRIPTION OF DRAWINGS

[0044] Figure 1 is a method flowchart of embodiment 1 of the present application;

[0045] Figure 2 is a system structure diagram of embodiment 2 of the present application. DETAILED DESCRIPTION

[0046] In order to better understand the above technical solutions, the above technical solutions will be described in detail below in combination with the drawings in the specification and specific embodiments.

[0047] The method provided by the present application can be implemented in a terminal environment, which can include one or more of the following components: a processor, a storage medium, and a display screen. The storage medium stores at least one instruction, which is loaded and executed by the processor to implement the method described in the following embodiments.

[0048] The processor can include one or more processing cores. The processor connects various parts in the entire terminal through various interfaces and lines, executes various functions of the terminal and processes data by running or executing instructions, programs, code sets or instruction sets stored in the storage medium, and calling data stored in the storage medium.

[0049] The storage medium can include random access memory (RAM) and read-only memory (ROM). The storage medium can be used to store instructions, programs, codes, code sets or instructions.

[0050] The display screen is used to display the user interface of each application program.

[0051] In addition, those skilled in the art can understand that the structure of the above-mentioned terminal does not constitute a limitation on the terminal, and the terminal can include more or fewer components, or combine certain components, or different component arrangements. For example, the terminal also includes radio frequency circuit, input unit, sensor, audio circuit, power supply and other components, which will not be described here.

[0052] Embodiment 1

[0053] As Figure 1As shown, the embodiment proposes an offline road crack detection method based on crack self-growth, comprising:

[0054] In step 101, a road image is acquired, the road image is segmented into a plurality of boundary-closed original regions, and the disturbance amount of texture change in each original region is calculated.

[0055] Specifically, the segmentation of the road image into a plurality of boundary-closed original regions comprises:

[0056] The road image is converted into a gray image.

[0057] The structural direction tensor matrix of each point in the gray image is extracted, the eigenvalue of the structural direction tensor matrix of each point is acquired, and the direction vector of each point is calculated through the eigenvalue. In the embodiment, the eigenvalue of the structural direction tensor matrix of each point and the calculation of the direction vector of each point are general technical means, and thus the embodiment will not be described in detail.

[0058] According to the structural direction tensor matrix and the direction vector, the principal direction direction vector of each point is calculated, and the principal direction direction angle is converted from the principal direction direction vector.

[0059] Specifically, the principal direction direction vector of each point is calculated by the following formula in the embodiment:

[0060]

[0061]

[0062] wherein, is the principal direction direction vector of a point with coordinates (x, y), is the direction vector of a point with coordinates (x, y), S(x, y) is the structural direction tensor matrix, T is a transpose operator, v′ x is the component of the direction vector of a point with coordinates (x, y) along the x-axis, v′ y is the component of the direction vector of a point with coordinates (x, y) along the y-axis.

[0063] Specifically, the principal direction direction angle is calculated by the following formula in the embodiment:

[0064]

[0065] wherein, θ(x, y) is the principal direction direction angle of a point with coordinates (x, y), v y is the principal direction direction vector of a point with coordinates (x, y) along the y-axis, v x is the principal direction direction vector of a point with coordinates (x, y) Component along the x-axis.

[0066] Calculate the local direction change rate of the main direction angle, which is used to describe the degree of mutation in the main direction. When the local direction change rate is greater than the direction change threshold, the current point is a structural change boundary. Traverse all points on the grayscale image to find all points on the structural change boundary. Perform connected region merging on the points on the non-structural change boundary, and each connected region is regarded as a piece of the original region.

[0067] Specifically, this embodiment calculates the local direction change rate using the following formula:

[0068]

[0069] Where Δθ(x, y) is the local direction change rate of the point with coordinates (x, y).

[0070] Step 102, judging whether each of the original regions is an activated region based on the disturbance amount, and identifying the crack starting point in each of the activated regions, wherein the remaining original regions are deleted;

[0071] Specifically, judging whether each of the original regions is an activated region according to the disturbance amount includes:

[0072] Calculating the pixel standard deviation of each original area, and deleting the original areas corresponding to the pixel standard deviations that are less than a preset standard deviation threshold, so as to avoid the occurrence of false cracks in shadow, pure white or pure black areas;

[0073] Obtain the perturbation vector of the remaining original area, which consists of the main direction perturbation index, the average edge response intensity and the structural skewness;

[0074] Specifically, this embodiment calculates the main direction disturbance index, average edge response strength, and structural skewness using the following formulas:

[0075] μ dir =std(θ(x,y)|(x,y)∈B k )

[0076] Among them, μ dir is the main direction disturbance indicator, std is the standard deviation symbol, B k is the kth remaining original region.

[0077]

[0078] Among them, μ edge is the average edge response strength, N is the kth remaining original area B k The number of internal pixels, I x is the grayscale gradient of the point with coordinates (x, y) along the x-axis, Iy is the gray gradient of the point with coordinates (x, y) along the y-axis.

[0079]

[0080] wherein μ skew is the structural skewness, n is the kth remaining original region B k is the number of main direction direction angles within B i is the kth remaining original region B k is the ith main direction direction angle within B is the kth remaining original region B k is the average of main direction direction angles within B

[0081] According to the disturbance vector, the total disturbance intensity of each remaining original region is calculated, and the original region corresponding to the total disturbance intensity of the remaining original region greater than the region disturbance intensity threshold is taken as an activated region.

[0082] Specifically, the total disturbance intensity is calculated by the following formula:

[0083]

[0084] wherein Φ k is the total disturbance intensity of the kth remaining original region B k .

[0085] Step 103, starting from the crack initiation point in the activated region, growing the crack trajectory according to the disturbance direction of the disturbance amount in the activated region until the crack trajectory growth is completed, thereby completing the road crack detection.

[0086] Specifically, expanding in the disturbance direction of the disturbance amount in the activated region to form a complete crack trajectory includes:

[0087] Taking an arbitrary point in the activated region, the total disturbance intensity of the arbitrary point is calculated, if the total disturbance intensity of the arbitrary point is greater than the point disturbance intensity threshold, the arbitrary point is taken as a crack initiation point, the main direction direction angle of the activated region is taken as the crack growth direction, and the crack trajectory growth starts from the crack initiation point;

[0088] Specifically, the total disturbance intensity of the arbitrary point is calculated by the following formula:

[0089]

[0090] wherein Φ'(x, y) is the total disturbance intensity of the point with coordinates (x, y), ρ dir (x, y) is the direction disturbance component of the point with coordinates (x, y), ρ edge(x, y) is an edge response component of a point with coordinates (x, y), and p skew (x, y) is a skewness component of a point with coordinates (x, y).

[0091] Specifically, the embodiment calculates each component by the following formula:

[0092]

[0093] (x, y) is an edge response component of a point with coordinates (x, y), and p dir (x, y) is a direction perturbation component of a point with coordinates (x, y), and p is the number of principal direction angles within the neighborhood is a 3x3 neighborhood of a point with coordinates (x, y), and p i′ is the number of principal direction angles within the neighborhood is the i'th principal direction angle within the neighborhood

[0094]

[0095] (x, y) is an edge response component of a point with coordinates (x, y), and p max is the maximum gray level gradient amplitude of all points in the gray level image, and p edge (x, y) is an edge response component of a point with coordinates (x, y).

[0096]

[0097] (x, y) is an edge response component of a point with coordinates (x, y), and p skew (x, y) is a skewness component of a point with coordinates (x, y). i′ is the number of principal direction angles within the neighborhood is the i'th principal direction angle within the neighborhood is the number of principal direction angles within the neighborhood

[0098] The step length is adopted to advance along the principal direction angle, and it is determined whether each new position satisfies the crack trajectory growth condition. If yes, the growth is continued. Otherwise, the point of the new position is taken as a crack end point, and a point within the neighborhood of the crack start point that satisfies the crack trajectory growth condition is found to continue the crack trajectory growth with the step length as the radius, wherein the crack trajectory growth condition is that the total perturbation intensity of the current point is recalculated at each new position. If the total perturbation intensity of the current point exceeds the point perturbation intensity threshold, the direction change amplitude of the current point is less than the direction change threshold, and the change amplitude of the total perturbation intensity of the current point is greater than the intensity change threshold, the growth is continued.

[0099] Specifically, the embodiment calculates the direction change amplitude of the current point by the following formula:

[0100] ​​Δθ' = min (|θ t - θ t-1 |, 2π - |θ t - θ t-1 |)

[0101] wherein, Δθ' is the change of the direction of the current point, θ t is the main direction of the current point at the fixed step t, and θ t-1 is the main direction of the corresponding point at the fixed step t-1.

[0102] Specifically, the embodiment calculates the change of the total disturbance intensity of the current point by the following formula:

[0103]

[0104] wherein, ΔΦ' is the change of the total disturbance intensity of the current point, Φ' t is the total disturbance intensity of the current point at the fixed step t, and Φ' t-1 is the total disturbance intensity of the corresponding point at the fixed step t-1.

[0105] Specifically, when growing the crack trajectory from the crack starting point, a fixed step is taken as the radius to find all points in the neighborhood of the crack starting point that satisfy the crack trajectory growth condition, and the crack trajectory growth is performed at the same time to find all branches of the crack trajectory until the crack ending point is found for each branch, and the complete crack trajectory growth is completed.

[0106] Embodiment 2

[0107] As shown in the following table, the embodiment proposes an offline road crack detection system based on crack self-growth, which comprises: Figure 2

[0108] A region division module is configured to acquire a road image, divide the road image into a plurality of original regions with closed boundaries, and calculate the disturbance amount of texture change in each original region.

[0109] Specifically, dividing the road image into a plurality of original regions with closed boundaries comprises:

[0110] Converting the road image into a gray image;

[0111] Extracting a structure direction tensor matrix of each point in the gray image, acquiring eigenvalues of the structure direction tensor matrix of each point, and calculating a direction vector of each point through the eigenvalues;

[0112] According to the structure direction tensor matrix and the direction vector, a main direction vector of each point is calculated, and the main direction vector is converted into a main direction angle.

[0113] ​A local direction change rate for describing a degree of mutation of the main direction is calculated for the main direction direction angle, and when the local direction change rate is greater than a direction change threshold value, the current point is a structure change boundary, all points on the gray scale image are traversed to find all points of the structure change boundary, and points of the non-structure change boundary are subjected to connected region merging, and each piece of connected region is taken as a piece of the original region.

[0114] A crack starting point identification module is configured to determine whether each of the original regions is an active region according to the disturbance amount, and identify a crack starting point in each of the active regions, wherein the remaining original regions are deleted.

[0115] Specifically, determining whether each of the original regions is an active region according to the disturbance amount comprises:

[0116] A pixel standard deviation of each of the original regions is calculated, and an original region corresponding to the pixel standard deviation less than a preset standard deviation threshold value is deleted to avoid a false crack in a shadow, a pure white region or a pure black region.

[0117] A disturbance vector of the remaining original regions is obtained, and the disturbance vector is composed of a main direction disturbance index, an average edge response intensity and a structure skewness.

[0118] A total disturbance intensity of the remaining original regions is calculated according to the disturbance vector, and an original region corresponding to the total disturbance intensity greater than a region disturbance intensity threshold value is taken as an active region.

[0119] A crack growth module is configured to grow a crack trajectory from a crack starting point in the active region in a disturbance direction of the disturbance amount in the active region until the crack trajectory growth is completed, so as to complete the road crack detection.

[0120] Specifically, expanding in the disturbance direction of the disturbance amount in the active region to form a complete crack trajectory comprises:

[0121] An arbitrary point in the active region is taken, and a total disturbance intensity of the arbitrary point is calculated, and if the total disturbance intensity of the arbitrary point is greater than a point disturbance intensity threshold value, the arbitrary point is taken as a crack starting point, a main direction direction angle of the active region is taken as a crack growth direction, and a crack trajectory is grown from the crack starting point;

[0122] The step length is adopted to advance along the main direction angle, it is judged whether each new position satisfies the crack trajectory growth condition, if yes, the growth is continued, otherwise, the point of the new position is taken as the crack end point, and the step length is taken as the radius to find the point in the crack start point neighborhood which satisfies the crack trajectory growth condition to continue the crack trajectory growth, wherein the crack trajectory growth condition is that the total disturbance intensity of the current point is recalculated at each new position, if the total disturbance intensity of the current point exceeds the point disturbance intensity threshold value, and the direction change amplitude of the current point is less than the direction change threshold value, and the change amplitude of the total disturbance intensity of the current point is greater than the intensity change threshold value, the growth is continued.

[0123] Specifically, when the crack trajectory growth is started from the crack start point, all the points in the crack start point neighborhood which satisfy the crack trajectory growth condition are found with the step length as the radius, and the crack trajectory growth is simultaneously carried out to find all the branches of the crack trajectory until the crack end point of each branch is found, and the complete crack trajectory growth is completed.

[0124] Embodiment 3

[0125] The embodiment of the present application also provides a storage medium which stores a plurality of instructions for realizing the off-line road crack detection method based on crack self-growth.

[0126] Optionally, in the embodiment, the storage medium can be located in any one of computer terminals in a computer terminal group in a computer network, or in any one of mobile terminals in a mobile terminal group.

[0127] Optionally, in the embodiment, the storage medium is configured to store program codes for executing the following steps: step 101, acquiring a road image, segmenting the road image into a plurality of boundary-closed original regions, and calculating disturbance amounts of texture changes in each original region;

[0128] Specifically, the step of segmenting the road image into a plurality of boundary-closed original regions comprises:

[0129] The road image is converted into a gray-scale image;

[0130] A structure direction tensor matrix of each point in the gray-scale image is extracted, eigenvalues of the structure direction tensor matrix of each point are acquired, and a direction vector of each point is calculated through the eigenvalues;

[0131] A main direction direction vector of each point is calculated according to the structure direction tensor matrix and the direction vector, and the main direction direction vector is converted into a main direction direction angle;

[0132] A local direction change rate for describing a degree of mutation of the main direction is calculated, and when the local direction change rate is greater than a direction change threshold value, the current point is a structure change boundary. All points of structure change boundaries are found by traversing all points on the gray image, and points of non-structure change boundaries are merged into connected regions, each of which is a block of the original region.

[0133] In step 102, it is judged whether each of the original regions is an active region according to the disturbance amount, and a crack starting point in each of the active regions is identified, wherein the remaining original regions are deleted.

[0134] Specifically, judging whether each of the original regions is an active region according to the disturbance amount comprises:

[0135] A pixel standard deviation of each of the original regions is calculated, and the original region corresponding to the pixel standard deviation less than a preset standard deviation threshold value is deleted to avoid a false crack in a shadow, a pure white or a pure black region.

[0136] A disturbance vector of the remaining original region is obtained, which is composed of a main direction disturbance index, an average edge response intensity and a structure skewness.

[0137] A total disturbance intensity of the remaining original region is calculated according to the disturbance vector, and the original region corresponding to the total disturbance intensity greater than a region disturbance intensity threshold value is taken as an active region.

[0138] In step 103, a crack trajectory is grown in the disturbance direction of the disturbance amount in the active region from the crack starting point in the active region until the crack trajectory is grown to completion, so that the road crack detection is completed.

[0139] Specifically, expanding in the disturbance direction of the disturbance amount in the active region to form a complete crack trajectory comprises:

[0140] An arbitrary point in the active region is taken, and a total disturbance intensity of the arbitrary point is calculated, and if the total disturbance intensity of the arbitrary point is greater than a point disturbance intensity threshold value, the arbitrary point is taken as a crack starting point, a main direction direction angle of the active region is taken as a crack growth direction, and a crack trajectory is grown from the crack starting point.

[0141] The step length is used to advance along the main direction angle, and it is determined whether each new position satisfies the crack trajectory growth condition. If yes, the growth is continued. Otherwise, the point of the new position is taken as the crack end point, and the step length is taken as the radius to find the point in the crack start point neighborhood that satisfies the crack trajectory growth condition to continue the crack trajectory growth. The crack trajectory growth condition is that the total disturbance intensity of the current point is recalculated at each new position. If the total disturbance intensity of the current point exceeds the point disturbance intensity threshold, the direction change amplitude of the current point is less than the direction change threshold, and the change amplitude of the total disturbance intensity of the current point is greater than the intensity change threshold, the growth is continued.

[0142] Specifically, when the crack trajectory growth is started from the crack start point, all the points in the crack start point neighborhood that satisfy the crack trajectory growth condition are found with the step length as the radius, and the crack trajectory growth is simultaneously performed to find all the branches of the crack trajectory until the crack end point of each branch is found, and the complete crack trajectory growth is completed.

[0143] Embodiment 4

[0144] The embodiment of the present application also provides an electronic device, which comprises a processor and a storage medium connected with the processor, and the storage medium stores a plurality of instructions, which can be loaded and executed by the processor to enable the processor to perform the offline road crack detection method based on crack self-growth.

[0145] Specifically, the electronic device of the embodiment can be a computer terminal, which can comprise one or more processors and a storage medium.

[0146] The storage medium can be used to store software programs and modules, such as the offline road crack detection method based on crack self-growth in the embodiment of the present application, and the corresponding program instructions / modules. The processor performs various functional applications and data processing by running the software programs and modules stored in the storage medium, that is, the offline road crack detection method based on crack self-growth is realized. The storage medium can comprise a high-speed random storage medium and can also comprise a non-volatile storage medium, such as one or more magnetic storage systems, flash memories, or other non-volatile solid-state storage media. In some examples, the storage medium can further comprise storage media remotely arranged relative to the processor, and these remote storage media can be connected to the terminal through a network. Examples of the network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0147] The processor can call the information and application programs stored in the storage medium through a transmission system to perform the following steps: step 101, acquiring a road image, segmenting the road image into a plurality of boundary-closed original regions, and calculating the disturbance amount of texture change in each original region;

[0148] Specifically, the road image is divided into a plurality of original regions with closed boundaries, including:

[0149] Converting the road image into a gray image;

[0150] Extracting a structure direction tensor matrix of each point in the gray image, obtaining an eigenvalue of the structure direction tensor matrix of each point, and calculating a direction vector of each point through the eigenvalue;

[0151] According to the structure direction tensor matrix and the direction vector, a principal direction direction vector of each point is calculated, and the principal direction direction vector is converted into a principal direction direction angle;

[0152] A local direction change rate of the principal direction direction angle for describing a degree of principal direction mutation is calculated, and when the local direction change rate is greater than a direction change threshold value, the current point is a structure change boundary. All points of the structure change boundary are found out by traversing all points in the gray image, and points of a non-structure change boundary are subjected to connected region merging. Each connected region is taken as an original region.

[0153] Step 102, judging whether each original region is an active region according to a disturbance amount, and identifying a crack starting point in each active region, wherein the remaining original regions are deleted;

[0154] Specifically, judging whether each original region is an active region according to a disturbance amount includes:

[0155] A pixel standard deviation of each original region is calculated, and the original region corresponding to the pixel standard deviation less than a preset standard deviation threshold value is deleted, so as to avoid a false crack in a shadow, a pure white region or a pure black region;

[0156] A disturbance vector of the remaining original region is obtained, which is composed of a principal direction disturbance index, an average edge response intensity and a structure skewness;

[0157] A total disturbance intensity of the remaining original region is calculated according to the disturbance vector, and the original region corresponding to the total disturbance intensity greater than a region disturbance intensity threshold value is taken as an active region.

[0158] Step 103, starting from the crack starting point in the active region, growing a crack trajectory in a disturbance direction of the disturbance amount in the active region until the crack trajectory is grown, so as to complete road crack detection.

[0159] Specifically, expanding in the disturbance direction of the disturbance amount in the active region to form a complete crack trajectory includes:

[0160] Taking an arbitrary point in the activation area, and calculating the total disturbance intensity of the arbitrary point, if the total disturbance intensity of the arbitrary point is greater than the point disturbance intensity threshold, the arbitrary point is taken as a crack starting point, a main direction direction angle of the activation area is taken as a crack growth direction, and a crack trajectory growth is started from the crack starting point;

[0161] A fixed step length is adopted to advance along the main direction direction angle, it is judged whether each new position satisfies a crack trajectory growth condition or not, if yes, the growth is continued, otherwise, a point of the new position is taken as a crack ending point, a fixed step length is taken as a radius, a point in a neighborhood of the crack starting point which satisfies the crack trajectory growth condition is found, and the crack trajectory growth is continued, wherein the crack trajectory growth condition is that the total disturbance intensity of the current point is recalculated at each new position, if the total disturbance intensity of the current point exceeds the point disturbance intensity threshold, a direction change amplitude of the current point is less than a direction change threshold, and a change amplitude of the total disturbance intensity of the current point is greater than an intensity change threshold, the growth is continued.

[0162] Specifically, when the crack trajectory growth is started from the crack starting point, all points in a neighborhood of the crack starting point which satisfy the crack trajectory growth condition are found with a fixed step length as a radius, and the crack trajectory growth is simultaneously performed, so as to find all branches of the crack trajectory, until a crack ending point of each branch is found, and the complete crack trajectory growth is performed.

[0163] The above-mentioned embodiment numbers of the present application are only for description, and do not represent the advantages and disadvantages of the embodiments.

[0164] In the above-mentioned embodiments of the present application, the description of each embodiment has its own emphasis, and the parts not described in detail in a certain embodiment can be referred to the relevant description of other embodiments.

[0165] In the several embodiments of the present application, it should be understood that the disclosed technology can be implemented in other ways. Of course, the system embodiments described above are only schematic. For example, the division of units is only a logical function division, and there can be another division manner in actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interface, unit or module, and can be electrical or other forms.

[0166] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e. they can be located in one place, or can be distributed on a plurality of network units. According to actual needs, part or all of the units can be selected to achieve the purpose of the present embodiment scheme.

[0167] In addition, each function unit in each embodiment of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software function unit.

[0168] When the integrated unit is realized in the form of a software function unit and sold or used as an independent product, it can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application, essentially or in the form of a contribution to the prior art, or all or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes several instructions for causing a computer machine (which can be a personal computer, a server, or a network machine, etc.) to execute all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: a U disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a mobile hard disk, a magnetic disk or an optical disk, and various media that can store program codes.

[0169] Obviously, the above embodiments are only examples for clear illustration, and are not a limitation on the implementation modes. For those skilled in the art, on the basis of the above description, other different forms of changes or variations can also be made. Here, it is not necessary and also impossible to enumerate all the implementation modes. The obvious changes or variations derived therefrom are still within the protection scope of the present application.

Claims

1. An off-line road crack detection method based on crack self-growth, characterized by, The method comprises the following steps: acquiring a road image, segmenting the road image into a plurality of boundary-closed original regions, and calculating a disturbance amount of texture change in each original region; judging whether each original region is an active region according to the disturbance amount, and identifying a crack starting point in each active region, wherein the remaining original regions are deleted; growing a crack track from the crack starting point in the active region in a disturbance direction of the disturbance amount in the active region until the crack track is grown completely, thereby completing road crack detection.

2. The off-line road crack detection method based on crack self-growth according to claim 1, wherein, The step of segmenting the road image into a plurality of boundary-closed original regions comprises the following steps: converting the road image into a gray image; extracting a structure direction tensor matrix of each point in the gray image, acquiring an eigenvalue of the structure direction tensor matrix of each point, and calculating a direction vector of each point through the eigenvalue; calculating a main direction direction vector of each point according to the structure direction tensor matrix and the direction vector, and converting the main direction direction vector into a main direction direction angle; calculating a local direction change rate of the main direction direction angle for describing a main direction mutation degree, when the local direction change rate is greater than a direction change threshold value, the current point is a structure change boundary, all points in the gray image are traversed to find all structure change boundary points, and non-structure change boundary points are subjected to connected region merging, each piece of connected region being taken as one original region.

3. The off-line road crack detection method based on crack self-growth according to claim 2, wherein, The step of judging whether each original region is an active region according to the disturbance amount comprises the following steps: calculating a pixel standard deviation of each original region, and deleting the original region corresponding to the pixel standard deviation less than a preset standard deviation threshold value, so as to avoid a false crack in a shadow, pure white or pure black region; acquiring a disturbance vector of the remaining original region, the disturbance vector being composed of a main direction disturbance index, an average edge response intensity and a structure skewness; calculating a total disturbance intensity of the remaining original region according to the disturbance vector, and taking the original region corresponding to the total disturbance intensity of the remaining original region greater than a region disturbance intensity threshold value as an active region.

4. The off-line road crack detection method based on crack self-growth according to claim 3, wherein, The step of expanding in the disturbance direction of the disturbance amount of the active region to form a complete crack track comprises the following steps: taking an arbitrary point in the active region, calculating a total disturbance intensity of the arbitrary point, if the total disturbance intensity of the arbitrary point is greater than a point disturbance intensity threshold value, the arbitrary point is taken as a crack starting point, a main direction direction angle of the active region is taken as a crack growth direction, and the crack track is grown from the crack starting point; a fixed step length is adopted to advance along the main direction direction angle, it is judged whether each new position satisfies a crack track growth condition, if yes, the growth is continued, otherwise, a point of the new position is taken as a crack ending point, a fixed step length is taken as a radius, and a point in a neighborhood of the crack starting point satisfying the crack track growth condition is found to continue the crack track growth, wherein the crack track growth condition is that the total disturbance intensity of the current point is recalculated at each new position, if the total disturbance intensity of the current point exceeds the point disturbance intensity threshold value, a direction change amplitude of the current point is less than a direction change threshold value, and a change amplitude of the total disturbance intensity of the current point is greater than an intensity change threshold value, the growth is continued.

5. The off-line road crack detection method based on crack self-growth according to claim 4, wherein, The crack trajectory growth is started from a crack starting point, a fixed step length is taken as a radius, all points in a neighborhood of the crack starting point satisfying the crack trajectory growth condition are found, and the crack trajectory growth is performed simultaneously to find all branches of the crack trajectory until each branch finds a crack ending point, and the complete crack trajectory growth is completed.

6. An off-line road crack detection system based on crack self-growth, characterized by, The method comprises the following steps: A region division module is configured to acquire a road image, divide the road image into a plurality of original regions with closed boundaries, and calculate a disturbance amount of texture variation in each original region. A crack starting point identification module is configured to determine whether each original region is an active region according to the disturbance amount, and identify a crack starting point in each active region, wherein the remaining original regions are deleted. A crack growth module is configured to grow a crack trajectory from the crack starting point in the active region in a disturbance direction of the disturbance amount in the active region until the crack trajectory growth is completed, thereby completing road crack detection.

7. An off-line road crack detection system based on crack self-growth as claimed in claim 6, wherein, The step of dividing the road image into a plurality of original regions with closed boundaries comprises the following steps: The road image is converted into a gray image. A structural direction tensor matrix of each point in the gray image is extracted, an eigenvalue of the structural direction tensor matrix of each point is acquired, and a direction vector of each point is calculated by using the eigenvalue. A main direction direction vector of each point is calculated according to the structural direction tensor matrix and the direction vector, the main direction direction vector is converted into a main direction direction angle, and a local direction change rate of the main direction direction angle for describing a main direction mutation degree is calculated. When the local direction change rate is greater than a direction change threshold value, the current point is a structure variation boundary, all points on the gray image are traversed to find all points of the structure variation boundary, and a connected region merging operation is performed on points that are not structure variation boundaries, and each connected region is taken as an original region.

8. An off-line road crack detection system based on crack self-growth as claimed in claim 7, wherein, The step of determining whether each original region is an active region according to the disturbance amount comprises the following steps: A pixel standard deviation of each original region is calculated, and an original region corresponding to the pixel standard deviation that is less than a preset standard deviation threshold value is deleted to avoid a false crack in a shadow, a pure white or a pure black region. A disturbance vector of a remaining original region is acquired, and the disturbance vector comprises a main direction disturbance index, an average edge response intensity and a structure skewness. A total disturbance intensity of the remaining original region is calculated according to the disturbance vector, and an original region corresponding to the total disturbance intensity of the remaining original region that is greater than a region disturbance intensity threshold value is taken as an active region.

9. An off-line road crack detection system based on crack self-growth as claimed in claim 8, wherein, The step of expanding in a disturbance direction of the disturbance amount in the active region to form a complete crack trajectory comprises the following steps: An arbitrary point in the active region is taken, a total disturbance intensity of the arbitrary point is calculated, if the total disturbance intensity of the arbitrary point is greater than a point disturbance intensity threshold value, the arbitrary point is taken as a crack starting point, a main direction direction angle of the active region is taken as a crack growth direction, and a crack trajectory growth is started from the crack starting point. The method comprises the following steps: adopting a fixed step length to advance along a main direction angle, judging whether each new position satisfies a crack trajectory growth condition, if yes, continuing to grow, otherwise, taking a point of the new position as a crack end point, taking the fixed step length as a radius, finding a point in a neighborhood of the crack start point satisfying the crack trajectory growth condition, and continuing to grow the crack trajectory, wherein the crack trajectory growth condition is that: recalculating a total disturbance intensity of a current point at each new position, if the total disturbance intensity of the current point exceeds a point disturbance intensity threshold value, a direction change amplitude of the current point is less than a direction change threshold value, and a change amplitude of the total disturbance intensity of the current point is greater than an intensity change threshold value, then continuing to grow.

10. An off-line road crack detection system based on crack self-growth as claimed in claim 9, wherein, When growing the crack trajectory from the crack start point, taking the fixed step length as a radius, finding all points in a neighborhood of the crack start point satisfying the crack trajectory growth condition, and simultaneously growing the crack trajectory, so as to find all branches of the crack trajectory, until each branch finds a crack end point, and the crack trajectory is completely grown.

Images