Road crack detection method based on connected region analysis

Abstract

The invention discloses a road crack detection method based on connected region analysis, and the method comprises the steps: obtaining a to-be-detected road image I, and carrying out the preprocessing of the to-be-detected road image I, and obtaining a binary image IB (x, y); detecting a connected region to obtain a connected region Rk; the fractal dimension Dk of each connected region Rk is calculated, whether the connected region Rk is a road gap is judged through the fractal dimension Dk, the road gap is removed, and a binary image with the road gap removed is obtained; traversing each connected region in the binary image from which the road gaps are removed, finding out the connected region with the largest area as a crack main region Rmain, and adding the crack main region Rmain into a crack region set Crack; the main direction angle theta of the crack main area Rmain is calculated, the step length of the sliding window is adjusted according to the main direction angle theta, the crack area is expanded by using the adjusted sliding window, and the crack area set Crack is updated; and obtaining a road crack region binary image according to the set Crack, and superposing the road crack region binary image into the original to-be-detected road image I to obtain a road crack detection result. The detection accuracy can be improved.

Description

Technical Field

[0001] The invention belongs to the technical field of image processing, and in particular relates to a road crack detection method based on connected region analysis. Background Art

[0002] With the increasing number of high-grade roads (such as expressways and urban expressways), the demand for road maintenance is also increasing. However, due to complex construction conditions, external environmental influences, and vehicle loads, quality problems and various defects such as cracks are unavoidable during the construction and use of roads. These defects will weaken the bearing capacity and service life of the road surface, and may also affect vehicle speed, fuel consumption, driving comfort and traffic safety.

[0003] At present, road inspection in my country mainly relies on manual work, usually completed through on-site measurement, recording, statistics and reporting. This method not only consumes a lot of manpower and material resources, but also has low inspection efficiency and insufficient accuracy, and also has high safety risks, especially when working on highways.

[0004] Therefore, developing an efficient and accurate method for automatic identification of road defects has become an urgent need in the field of road inspection and maintenance. Intelligent equipment provides a technical basis for road data collection, but professional computer algorithms and software are still needed for defect identification, classification, reporting and review of video data. By automating road defect detection, it can not only improve the work efficiency of relevant departments, but also provide a scientific decision-making basis for road management and maintenance, effectively reducing manpower and time costs. Summary of the invention

[0005] In order to solve the above technical problems, the present invention proposes a road crack detection method based on connected region analysis, which comprises:

[0006] S1: Obtain the road image I to be detected, and obtain the binary image I after preprocessing B (x,y);

[0007] S2: Binarized image I B After the connected region detection is performed on (x, y), the adjacent foreground pixels are aggregated into connected regions and marked to obtain the marked connected region R k ;

[0008] S3: Calculate the connected area R after each label k The fractal dimension D k , through the fractal dimension D k Determine the connected area R k Whether it is a road gap, remove the road gap to obtain a binary image without the road gap;

[0009] S4: Traverse each connected region in the binary image after removing the road gaps, and find the connected region with the largest area as the main crack region R main , add it to the crack region set C crack middle;

[0010] S5: Calculate the main crack area R main The main direction angle θ is used to adjust the sliding window step size according to the main direction angle θ, and the adjusted sliding window is used to expand the crack area and update the crack area set C crack ;

[0011] S6: Based on the updated crack area set C crack A binary image of the road crack area is obtained, which is superimposed on the original road image I to be detected to obtain a road crack detection result.

[0012] Furthermore, the connected area R in the binary image without the road gap is calculated. k Dynamic threshold of area T P and dynamic threshold W of the bounding rectangle width P’ , through the area dynamic threshold T P and dynamic threshold W of the bounding rectangle width P’ From each connected region R k The main body of road cracks can be accurately identified.

[0013] Beneficial effects of the present invention: The present invention first performs Gaussian smoothing on the original image and calculates the image threshold, and divides the original image into a binary image according to the threshold; then, the connected area of ​​the binary image is marked by connected area detection; the fractal dimension of each connected area is calculated by the box counting method, and effective road cracks are identified based on the complexity of the fractal dimension; then, the minimum width and area threshold of the circumscribed rectangle of the connected area are calculated to remove noise with smaller morphology, and finally, the crack area is expanded by sliding window and angle expansion, thereby further removing noise and obtaining the final crack area. The method effectively reduces misjudgment and missed judgment, and improves the accuracy and stability of road crack detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 This is a typical road crack image;

[0015] Figure 2 is a flowchart of steps of an embodiment of the present invention;

[0016] Figure 3 is a road image to be detected collected in an embodiment of the present invention;

[0017] Figure 4 is a binary image of the road image to be detected in the embodiment of the present invention after preprocessing;

[0018] Figure 5 It is a binary image obtained by removing road gaps through fractal dimension in an embodiment of the present invention;

[0019] Figure 6 A binary image with noise removed in an embodiment of the present invention;

[0020] Figure 7 It is a binary image after the crack area is expanded and the noise is further removed in the embodiment of the present invention;

[0021] Figure 8 This is a diagram of the final crack detection results in an embodiment of the present invention. DETAILED DESCRIPTION

[0022] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0023] Intelligent equipment provides a technical basis for collecting road data (including road crack images).

[0024] Figure 1 A typical road crack image.

[0025] The road crack image collection vehicle takes typical road crack images, referring to Figure 1 As shown in the figure, the main features of the road crack image include: (1) the image has uneven illumination. The upper left corner of the image is brighter and the lower right corner is darker, resulting in changes in the contrast of the cracks in different areas; (2) the background texture of the image is complex. There are many textures and small spots on the road surface in the image. These textures may be misjudged as cracks, especially when the grayscale difference is not obvious, which increases the detection noise; (3) the crack edges in the image are blurred. The edges of the road cracks in the image are not clear enough and there is some slight blur, and the crack contrast is low; the contrast of some crack areas is low and may be close to the grayscale value of the surrounding background. The above features increase the difficulty of crack detection. How to reduce the impact of uneven illumination and complex background texture on the crack body and detection accuracy is an urgent problem to be solved.

[0026] The embodiment of the present invention proposes a road crack detection method based on connected region analysis, such as Figure 2 As shown, the method includes:

[0027] S1: Obtain the road image I to be detected, and obtain the binary image I after preprocessing B (x,y).

[0028] Figure 3 It is the road image to be detected collected in the embodiment of the present invention.

[0029] Specifically, the road image I to be detected is collected by intelligent equipment, referring to Figure 3 The pixel value of the road image I to be detected is represented as I(x, y), where (x, y) represents the pixel coordinates of the image I, x represents the horizontal coordinate, and y represents the vertical coordinate.

[0030] Preferably, the preprocessing includes Gaussian blur processing, smoothing processing and binarization processing, and the binarization processing includes: calculating an adaptive threshold T(x, y), converting the pixel value I of the image after Gaussian blur processing to G (x, y) is compared with the adaptive threshold T(x, y) to generate a binary image I B (x,y).

[0031] The collected road image I to be detected is subjected to Gaussian blur and smoothing processing respectively, and the pixel value I after Gaussian blur processing is G (x,y) is specifically:

[0032]

[0033] Among them, I G (x, y) represents the pixel value after Gaussian blur processing is performed on the road image I to be detected, G(i, j) represents the Gaussian kernel, i represents the offset relative to x, j represents the offset relative to y, σ is the standard deviation, which determines the degree of blur, e represents the natural logarithm, k represents half the size of the sliding window of the Gaussian blur kernel, and the size of the sliding window is 2k+1. For example, when k=2, the window size is 5.

[0034] The adaptive threshold segmentation method for the image after Gaussian blur processing separates the background image from the image. The adaptive threshold dynamically calculates the threshold according to the local environment of the pixel, so that different illumination areas in the image can be reasonably segmented. The adaptive threshold T(x,y) is specifically:

[0035]

[0036] Where N is the number of pixels in the window, and C is a constant offset used to adjust the sensitivity of the segmentation. For example, if the size of the local neighborhood window is 11×11, then the number of pixels in the window N=121, and the constant offset C=2.

[0037] The pixel value I of the image after Gaussian blur processing G (x, y) is compared with the adaptive threshold T(x, y) to generate a binary image I B(x,y), which is specifically:

[0038]

[0039] Among them, when I G When (x,y)<T(x,y), let I B (x, y) = 255, the pixel is white; in other cases (i.e. I G (x,y)≥T(x,y)), let I B (x, y) = 0, the pixel appears black.

[0040] For example, the road crack area or the road gap area is set to white and the background is set to black.

[0041] Figure 4 It is a binary image after preprocessing of the road image to be detected in the embodiment of the present invention.

[0042] Reference Figure 4 As shown, in the binary image I B In (x, y), the crack area will form an obvious white area, but there are smaller noises and a large number of them. The shapes and sizes of the noises are different, and the noise distribution is uneven, mainly concentrated on the left side of the image, with fewer noise points on the right side.

[0043] S2: Binarized image I B After the connected region detection is performed on (x, y), the adjacent foreground pixels are aggregated into connected regions and marked to obtain the marked connected region R k and the total number of connected regions n.

[0044] The connected region R k It is expressed as:

[0045]

[0046] Among them, R k represents the kth connected area, N(x,y) represents the neighborhood of the pixel point (x,y), and (x',y') represents the pixel point in the neighborhood of the pixel point (x,y).

[0047] The connected areas are marked to determine each connected area, a marked graph L(x,y) is generated, and a unique area number is assigned to each connected area, which is expressed as:

[0048] L(x,y)=k,if(x,y)∈R k ,

[0049] Among them, k represents the connected area number, and the total number of connected areas is n.

[0050] S3: Calculate the connected area R after each label k The fractal dimension D k , through the fractal dimension D k Determine the connected area R k Whether it is a road gap, the road gap is eliminated to obtain a binary image without the road gap.

[0051] The fractal dimension D k The smaller it is, the more regular the connected area is. Since the morphology of the crack area is more complex, its fractal dimension is larger, while the road gap is more regular and has a smaller fractal dimension, a threshold is set to delete the area with a smaller fractal dimension.

[0052] Preferably, the box counting method is used to calculate the connected area R after each mark k The fractal dimension D k .

[0053] Box counting is an empirical technique for estimating the fractal dimension of an object, image, or set. The basic idea is to cover the object with smaller and smaller boxes and then count the number of boxes needed to cover the object at each corresponding scale. The measurement obtained in this way is called the "box counting dimension" or "Minkowski dimension". If an object is fractal in nature, it should have similar structures or patterns at different scales.

[0054] The box counting process includes:

[0055] S301: Select a box size, a gradually decreasing box size.

[0056] Specifically, define a set of scales S, i.e., the side length of the box, for example, s = [2, 4, 8, 16, 32, 64]. Each scale S divides the image into a grid of size S × S. For each scale S, calculate the number of boxes N that contain the target pixel. S :

[0057]

[0058] Among them, 1(·) is the indicator function, when S ij When (s)>0, the value is 1, otherwise it is 0. M(s) and N(s) are the number of rows and columns of the divided image, respectively, which are:

[0059]

[0060] Where H and W are the height and width of the original image, S ij (s) is the cumulative sum of the pixel values ​​in the grid (x, y), which is specifically:

[0061]

[0062] Repeat the above steps to get the number of non-empty boxes at different scales S:

[0063]

[0064] S302: Cover the object: Cover the object with boxes of each size and record the number of boxes N required s .

[0065] S303: Draw a log-log graph: Draw a log-log graph with the logarithm of the box size S as the horizontal axis and the logarithm of the required number of boxes N as the vertical axis.

[0066] Specifically, for the box size S and the number of boxes N s Take the logarithm X = log(s), Y = log(N s ), we get a linear relationship (i.e., a log-log plot), which is:

[0067] Y=-D·X+b (7)

[0068] Among them, the slope -D represents the fractal dimension, and the intercept b represents the offset, which is independent of the fractal dimension.

[0069] The least squares formula is used to fit the logarithmic relationship, specifically:

[0070]

[0071] Preferably, the threshold D is set t , the fractal dimension D k With threshold D t For comparison, if D k <D t Then determine the connected area R k is the road gap and removes it. If D k ≥D t Then determine the connected area R k Not a road gap and keep it. Example, D t =1.1.

[0072] Figure 5 is a binary image obtained by removing the road gaps through fractal dimension in the embodiment of the present invention. Figure 5 In the figure, the relatively regular road gap connected area on the left is deleted, and the crack body and irregular noise are retained. It is still necessary to further remove the irregular noise in the binary image after removing the road gap.

[0073] Preferably, the connected area R in the binary image after removing the road gap is calculated. k Dynamic threshold of area TP and dynamic threshold W of the bounding rectangle width P’ , through the area dynamic threshold T P and dynamic threshold W of the bounding rectangle width P’ From each connected region R k The main body of road cracks can be accurately identified.

[0074] Specifically, in each connected region R in the binary image after removing the road gaps i , calculate its area A i , that is, the number of all pixels in the area.

[0075] The area calculation formula is:

[0076]

[0077] Collect the area values ​​of all connected regions in the binary image without road gaps into a set n is the total number of connected areas. Set the quantile P and calculate the quantile value T of the area P , and its calculation formula is:

[0078]

[0079] Among them, T P Indicates the dynamic threshold of area, which indicates the area value under this quantile.

[0080] For each connected region R in the binary image after removing the road gaps i , find its width W through the enclosing rectangle i , and its calculation formula is:

[0081] W i =x max -x min (11)

[0082] Among them, x max Represents the connected region R i The maximum coordinate value in the horizontal direction, x min Represents the connected region R i The minimum coordinate value in the horizontal direction.

[0083] The width W of all connected regions in the binary image after removing the road gaps i Composition Collection n is the total number of connected areas. Set the quantile P' and calculate the quantile value W of the circumscribed rectangle P′ , and its calculation formula is:

[0084]

[0085] Among them, W P′ Indicates the dynamic threshold of the enclosing rectangle width, which indicates the enclosing rectangle width value under this quantile.

[0086] In this example, set P = 50%, P' = 90%, and calculate T P and W P′ , traverse all connected regions R in the binary image with road gaps removed, and only retain connected regions that meet the following conditions:

[0087] R={R i |A(R i )≥T P andW(R i )≥W P′}.

[0088] Figure 6 It is a binary image with noise removed in the embodiment of the present invention.

[0089] Reference Figure 6 As shown, the dynamic threshold T of the area P and dynamic threshold W of the bounding rectangle width P’ The combination of the two further effectively removes most of the noise and false detection areas, but still retains a few large noise areas. However, the main body of the road cracks still needs to be identified more accurately.

[0090] S4: Traverse each connected region in the binary image after removing the road gap, find the connected region with the largest area and take it as the main crack region R main , add it to the crack region set C crack middle.

[0091] Specifically: Initialize the set of crack regions, traverse each connected region in the binary image without road gaps, and find the connected region with the largest area as the main crack region R main , add it to the crack region set C crack middle.

[0092] The connected region R with the largest area main The calculation formula is:

[0093]

[0094] Among them, R i Represents the i-th connected region in the binary image after removing the road gaps.

[0095] At this time, the set of crack regions is expressed as: C crack = {R main}.

[0096] S5: Calculate the main crack area R main The main direction angle θ is used to adjust the sliding window step size according to the main direction angle θ, and the adjusted sliding window is used to traverse each connected area to update the crack area set C crack .

[0097] Preferably, the second-order moment is used to calculate the main crack area R main The main direction angle θ.

[0098] In the field of image processing technology, image moments refer to the weighted average (moment) of the grayscale of certain specific pixels in an image, or attributes of an image with similar functions or meanings. Image moments are often used to describe segmented image objects. Some properties of an image can be obtained through image moments, including area (or overall brightness), as well as information about the geometric center and direction.

[0099] For the main crack area R main , whose p+q-order moment (sometimes called the “raw moment”) is expressed as:

[0100]

[0101] Among them, M pq Represents the main crack area R main The p+q moment, x represents the horizontal coordinate of the pixel, y represents the vertical coordinate of the pixel, R main (x,y) represents the pixel value of the main crack area.

[0102] For the main crack area R main , its second-order moments include: horizontal second-order moment M 20 , vertical second moment M 02 and the joint second moment M 11 . Specifically:

[0103] Horizontal second moment M 20 (p=2, q=0) reflects the degree of expansion in the horizontal direction.

[0104] Vertical second moment M 02 (p=0, q=2) reflects the degree of expansion in the vertical direction.

[0105] Joint second moment M 11 (p=1, q=1) reflects the joint expansion relationship in the horizontal and vertical directions.

[0106] Through the crack main area R main The second moment M 20 ,M 02 ,M 11 , calculate its main direction angle θ as:

[0107]

[0108] Preferably, the sliding window step size is adjusted according to the main direction angle θ, and the adjusted sliding window is used to traverse each connected area to update the crack area set C crack The process is as follows: initialize the size of the sliding window w, calculate the step size of the sliding window according to the main direction angle θ; use the sliding window to select the crack area set C crack Starting from the left and right endpoints in the figure, expand the crack area along the main direction and update the crack area set C crack . Specifically:

[0109] S501: Initialize the size w of the sliding window, and calculate the step size of the sliding window according to the main direction angle θ.

[0110] Specifically, the horizontal step size S of the sliding window is calculated according to the main direction angle θ x and vertical step length S y , which are specifically:

[0111]

[0112] Among them, w represents the sliding window size.

[0113] When initializing the sliding window size, w=30.

[0114] In the crack region set Ccrack, the coordinate of its left endpoint is (x left ,y left ), the coordinate of its right endpoint is (x right ,y right ), the left and right endpoint coordinates are obtained from the following formula:

[0115]

[0116] Among them, x left Represents the set of crack regions C crack The horizontal coordinate of the left endpoint, y left Represents the set of crack regions C crack The ordinate of the left endpoint, x right Represents the set of crack regions C crack The horizontal coordinate of the right endpoint, y right Represents the set of crack regions C crack The ordinate of the left endpoint.

[0117] S502: Using a sliding window, from the crack region set C crack Starting from the left and right endpoints in the figure, expand the crack area along the main direction and update the crack area set C crack .

[0118] At each step of expansion, the pixels in the window are extracted, and the connected area pixel set that meets the conditions is added to the crack area set C crack middle.

[0119] Assume that the current extension endpoint is (x current ,y current ), the size of the window is w, and the range of the window is:

[0120]

[0121] If extended along the main direction (right endpoint), it is specifically:

[0122] (xcurrent,ycurrent)=(xcurrent+Sx,ycurrent+Sy);

[0123] If extended in the opposite direction (left endpoint), it is specifically:

[0124] (xcurrent,ycurrent)=(xcurrent-Sx,ycurrent-Sy);

[0125] Use the sliding window to expand the crack area. When it reaches the image boundary or there is no connected area in the window, it stops and updates the crack area set C. crack .

[0126] In each expansion step, from the current endpoint position (x current ,y current ) to generate a window. Let the set of all pixels in the window be W, and the pixels of the current crack set in the window be W crack , the label set of all connected regions in the window is W labels .

[0127] Preferably, the expansion condition of the crack region and the set of crack regions C crack The specific update rules are:

[0128] Find the new connected region pixels and their new connected regions that are not in the current crack set within the sliding window W;

[0129] For each new connected region, if the pixels it contains are adjacent to the existing regions in the current crack set, the entire new connected region is added to the crack set C. crack Update Crack Set C crack The left and right endpoints of ;

[0130] Using the sliding window, from the updated crack set C crack Starting from the left and right endpoints, expand the crack area and update the crack area set C crack , until the stopping condition is met. Specifically:

[0131] S511 Condition Detection: Find a crack that is not in the current crack set C in the sliding window W. crack The new connected area pixels in the , that is, the pixel set P that meets the conditions new :

[0132]

[0133] S512 New region check: Find the connected region to which the new connected region pixel belongs is R new , that is, the connected region labels to which these new pixels belong: R new = {L(x,y)|(x,y)∈P new}.

[0134] S513 Region connection detection: For each new connected region R i ∈R new , if the pixels it contains are adjacent to the area already in the current crack set, the entire area is added to the crack set.

[0135] The following conditions can be used to judge:

[0136]

[0137] Then R i All pixels are added to C crack In collection.

[0138] S514 Update the crack area set C crack And its left and right endpoints:

[0139] The new connected region R that meets the conditions i All pixels are added to the crack region set C crack :C crack =C crack ∪R i ; At this time, the left and right endpoint coordinates of the new crack area set are updated to:

[0140]

[0141] Continuously add new areas that meet the conditions to the crack set C crack It stops when it expands to the image boundary or there are no new connected areas in the window.

[0142] Figure 7 This is a binary image after the crack area is expanded and the noise is further removed in the embodiment of the present invention.

[0143] After the crack expansion stops, the connected areas that do not enter the set are regarded as noise. After deleting them, the complete crack image is obtained as shown in the attached figure. Figure 7 As shown. Figure 7 In the figure, it can be seen that after expanding the crack area and further removing the noise, the main area of ​​the crack is well retained, and the connected area that does not enter the crack set is successfully deleted.

[0144] S6: Based on the updated crack area set C crack A binary image of the road crack area is obtained, which is superimposed on the original road image I to be detected to obtain a road crack detection result.

[0145] Figure 8 This is a diagram of the final crack detection results in an embodiment of the present invention.

[0146] Experimental verification: The method of the present invention was used to detect a verification set (including 100 labeled road crack images). By comparing the detection results with the manually labeled real crack areas, the detected cracks covered about 80% of the real crack areas, and most of the key areas were successfully detected.

[0147] A person skilled in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the relevant hardware through a program, and the program can be stored in a computer-readable storage medium, which can include: ROM, RAM, disk or CD, etc.

[0148] 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 road crack detection method based on connected region analysis, characterized in that: include: Get the road image I to be detected, and get the binary image I after preprocessing B (x,y); For the binary image I B After the connected region detection is performed on (x, y), the adjacent foreground pixels are aggregated into connected regions and marked to obtain the marked connected region R k ; Calculate the connected area R after each label k The fractal dimension D k , through the fractal dimension D k Determine the connected area R k Whether it is a road gap, remove the road gap to obtain a binary image without the road gap; Traverse each connected area in the binary image without road gaps, find the connected area with the largest area and take it as the main crack area R main , add it to the crack region set C crack middle; Calculate the main crack area R main The main direction angle θ is used to adjust the sliding window step size according to the main direction angle θ, and the adjusted sliding window is used to expand the crack area and update the crack area set C crack ; According to the updated crack area set C crack A binary image of the road crack area is obtained, which is superimposed on the original road image I to be detected to obtain a road crack detection result.

2. The road crack detection method based on connected region analysis according to claim 1, characterized in that: The preprocessing includes Gaussian blur processing, smoothing processing and binarization processing. The binarization processing includes: calculating an adaptive threshold T(x, y), converting the pixel value I of the image after Gaussian blur processing G (x, y) is compared with the adaptive threshold T(x, y) to generate a binary image I B (x,y).

3. The road crack detection method based on connected region analysis according to claim 1, characterized in that: The box counting method is used to calculate the connected area R after each label k The fractal dimension D k .

4. The road crack detection method based on connected region analysis according to claim 1 or 3, characterized in that: Set the threshold D t , the fractal dimension D k With threshold D t For comparison, if D k <D t Then determine the connected area R k is the road gap and removes it. If D k ≥D t Then determine the connected area R k Not a road gap and keep it that way.

5. The road crack detection method based on connected region analysis according to claim 1 or 4, characterized in that: Calculate the connected area R in the binary image after removing the road gap k Dynamic threshold of area T P and dynamic threshold W of the bounding rectangle width P’ , through the area dynamic threshold T P and dynamic threshold W of the bounding rectangle width P’ From each connected region R k The main body of road cracks can be accurately identified.

6. The road crack detection method based on connected region analysis according to claim 1, characterized in that: The main crack area R is calculated using the second-order moment main The main direction angle θ.

7. The road crack detection method based on connected region analysis according to claim 1, characterized in that: Adjust the sliding window step size according to the main direction angle θ, use the adjusted sliding window to expand the crack area, and update the crack area set C crack , the process is: Initialize the size w of the sliding window and calculate the step size of the sliding window according to the main direction angle θ; Using the sliding window, from the crack area set C crack Starting from the left and right endpoints in the figure, expand the crack area along the main direction and update the crack area set C crack .

8. The road crack detection method based on connected region analysis according to claim 7, characterized in that: The expansion condition of the crack region and the set of crack regions C crack The specific update rules are: Find the new connected region pixels and their new connected regions that are not in the current crack set within the sliding window W; For each new connected region, if the pixels it contains are adjacent to the existing regions in the current crack set, the entire new connected region is added to the crack set C. crack Update Crack Set C crack The left and right endpoints of ; Using the sliding window, from the updated crack set C crack Starting from the left and right endpoints, expand the crack area and update the crack area set C crack , until the stopping condition is met.

Images