Bridge deck quality monitoring method and system based on edge detection

Abstract

The invention relates to the technical field of image processing, in particular to a bridge deck quality monitoring method and system based on edge detection, and the method comprises the steps: obtaining a bridge gray-scale map; obtaining a plurality of bridge edge maps through threshold iteration based on the bridge grey-scale map; according to the change of edge pixel points in adjacent bridge edge images, obtaining the possibility that suspected crack edges in all bridge edge images are crack edges, and further obtaining a plurality of reference edges and to-be-identified pixel points; obtaining a suspected splicing edge of each reference edge; according to the shape of each reference edge and the shape of the suspected spliced edge, obtaining the spliced edge; and taking all the spliced edges and the reference edge as a final crack edge in the bridge. According to the method, the crack edge is accurately identified by analyzing the similar extension direction characteristics of the crack edge under different thresholds and the matching characteristics of the two edges of the crack, so that the accuracy of bridge deck quality monitoring is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a bridge deck quality monitoring method and system based on edge detection. Background Art

[0002] Cracks in bridges may affect the structural safety and bearing capacity of bridges, increase the risk of corrosion, cause deterioration of steel bars and concrete, and endanger traffic safety. Therefore, it is necessary to detect cracks in bridges in a timely manner in order to formulate reasonable maintenance and reinforcement plans and extend the service life of bridges. However, in the traditional process of identifying crack edges through machine vision, it is necessary to collect images of the bridge surface. When collecting images of the bridge surface, it will be affected by external light, resulting in the inability of traditional edge detection operators to accurately identify the crack edges in bridges. Summary of the invention

[0003] The present invention provides a bridge deck quality monitoring method and system based on edge detection to solve the existing problem that the traditional edge detection operator cannot accurately identify the crack edges in the bridge.

[0004] The bridge deck quality monitoring method and system based on edge detection of the present invention adopts the following technical solutions:

[0005] An embodiment of the present invention provides a bridge deck quality monitoring method based on edge detection, the method comprising the following steps:

[0006] Get the grayscale image of the bridge;

[0007] Based on the bridge grayscale image, several bridge edge images are obtained through threshold iteration; according to the gradient value of the edge pixel points in the bridge edge image, some crack edge pixel points are obtained; according to the changes of the edge pixel points in the adjacent bridge edge images, the suspected crack edges in all bridge edge images are obtained; according to the suspected crack edges and some crack edge pixel points in all bridge edge images, the possibility that the suspected crack edges in all bridge edge images are crack edges is obtained; according to the possibility that the suspected crack edges in all bridge edge images are crack edges, several reference edges and pixels to be identified are obtained, and the pixels in the reference edges are reference pixels;

[0008] According to the gradient direction of the reference pixel points in the same reference edge, the standard crack width of each reference edge is obtained; according to the reference edge and its standard crack width, the suspected splicing edge of each reference edge is obtained; according to the shapes of the reference edge and its suspected splicing edge, the similarity between each suspected splicing edge and the corresponding reference edge is obtained;

[0009] According to the similarity between the suspected joint edge and the corresponding reference edge, the joint edge is obtained; all the joint edges and the reference edges are taken as the final crack edges in the bridge.

[0010] Preferably, the method of obtaining a plurality of bridge edge images by threshold iteration based on the bridge grayscale image includes the following specific methods:

[0011] Preset an initial threshold α, a termination threshold β, and a threshold iteration step length γ; set the threshold of the edge detection operator to α, use the edge detection operator to detect the edge in the bridge grayscale image, and obtain a first bridge edge image;

[0012] The threshold of the edge detection operator is set to α-γ, and the edge detection operator is used to detect the edge in the bridge grayscale image to obtain a second bridge edge image;

[0013] Until the threshold of the edge detection operator is set to β, the edge detection operator is used to detect the edge in the bridge grayscale image to obtain the xth bridge edge image, where

[0014] Preferably, the method of obtaining some crack edge pixel points according to the gradient values ​​of edge pixel points in the bridge edge map includes the following specific methods:

[0015] A crack edge gradient threshold coefficient δ is preset, the maximum gradient value in the first bridge edge map is obtained, and the pixel points in the first bridge edge map whose gradient values ​​are greater than δ times the maximum gradient value are taken as crack edge pixels.

[0016] Preferably, the method of obtaining the suspected crack edges in all bridge edge maps according to the changes of edge pixels in adjacent bridge edge maps includes:

[0017] For any edge pixel point in the second bridge edge map, the edge pixel point is recorded as the target point, and the pixel point in the first bridge edge map with the same position coordinates as the target point is recorded as the corresponding point of the target point. If the corresponding point is a crack edge pixel point, the target point is the crack edge pixel point. If the corresponding point is not a crack edge pixel point, the target point is a suspected crack edge pixel point. The adjacent suspected crack edge pixel points in the eight neighborhoods are classified as the same suspected crack edge to obtain several suspected crack edges; the adjacent crack edge pixel points in the eight neighborhoods are classified as the same crack edge to obtain several crack edges.

[0018] Preferably, the method of obtaining the possibility that the suspected crack edges in all bridge edge maps are crack edges according to the suspected crack edges and some crack edge pixels in all bridge edge maps includes:

[0019] For any suspected crack edge, the crack edge closest to the suspected crack edge is obtained and recorded as the target edge; all pixels between the suspected crack edge and the target edge are recorded as feature points, and the possibility of the suspected crack edge being a crack edge is obtained according to the gradient direction of the pixel points corresponding to all feature points in the bridge grayscale image and the distance between the suspected crack edge and the target edge. The specific calculation formula is:

[0020] P = exp[-(σ×L)]

[0021] Wherein, P represents the possibility that the suspected crack edge is a crack edge; σ represents the standard deviation of the gradient direction of the pixel points corresponding to all feature points in the bridge grayscale image; L represents the distance between the suspected crack edge and the target edge; exp[] represents an exponential function with a natural constant as the base.

[0022] Preferably, the method of obtaining a plurality of reference edges and pixels to be identified according to the possibility that the suspected crack edges in all bridge edge images are crack edges includes the following specific methods:

[0023] A possibility threshold μ of the suspected crack edge being a crack edge is preset. If the possibility that the suspected crack edge is a crack edge is greater than or equal to the possibility threshold, the suspected crack edge is a crack edge; if the possibility that the suspected crack edge is a crack edge is less than the possibility threshold, the suspected crack edge is not a crack edge; all pixel points in the crack edge are regarded as crack edge pixel points;

[0024] For any edge pixel point in the third bridge edge map, the edge pixel point is recorded as a target point, and the pixel point with the same position coordinates as the target point in the second bridge edge map is recorded as a corresponding point. If the corresponding point is a crack edge pixel point, the target point is a crack edge pixel point. If the corresponding point is not a crack edge pixel point, the target point is a suspected crack edge pixel point. The adjacent suspected crack edge pixel points in the eight neighborhoods are classified as the same suspected crack edge to obtain several suspected crack edges; the adjacent crack edge pixel points in the eight neighborhoods are classified as the same crack edge to obtain several crack edges.

[0025] The possibility of the suspected crack edge in the third bridge edge map being a crack edge is calculated, thereby obtaining several crack edges;

[0026] By analogy, all crack edges in the x-th bridge edge map are obtained, the crack edges are recorded as reference edges, the pixels on the reference edges are recorded as reference pixels, and all edge pixels in the x-th bridge edge map that are not reference pixels are recorded as pixels to be identified.

[0027] Preferably, the method of obtaining the suspected joint edge of each reference edge according to the reference edge and its standard crack width includes:

[0028] For any reference edge, a magnification coefficient τ is preset; all pixel points to be identified whose distance from the reference edge is less than τ times its standard crack width are recorded as suspected stitching pixel points of the reference edge, and the suspected stitching pixel points adjacent to the eight neighborhoods are classified as the same suspected stitching edge, so as to obtain several suspected stitching edges of the reference edge.

[0029] Preferably, the obtaining of the similarity between each suspected splicing edge and the corresponding reference edge according to the shapes of the reference edge and the suspected splicing edge includes the following specific methods:

[0030] For any suspected stitching edge of any reference edge; perform DTW matching on the reference edge and the suspected stitching edge to obtain several matching pairs; obtain the similarity between the suspected stitching edge and the reference edge according to the difference between the gradient directions of two pixels in each matching pair, and the specific calculation formula is:

[0031]

[0032] Where S represents the similarity between the suspected stitching edge and the reference edge; m represents the number of matching pairs; θ i,1,2 represents the angle between the first pixel and the second pixel in the i-th matching pair in the gradient direction; cos() represents the cosine trigonometric function; || represents the absolute value function; exp[] represents the exponential function with a natural constant as the base.

[0033] Preferably, the step of obtaining the stitching edge according to the similarity between the suspected stitching edge and the corresponding reference edge includes the following specific methods:

[0034] Preset a similarity threshold For any suspected stitching edge of any reference edge, if the similarity between the suspected stitching edge and the reference edge is greater than or equal to Then the suspected stitching edge is the stitching edge.

[0035] Another embodiment of the present invention provides a bridge deck quality monitoring system based on edge detection, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements any step of the above-mentioned bridge deck quality monitoring method based on edge detection when executing the computer program.

[0036] The beneficial effects of the technical solution of the present invention are as follows: the present application obtains several bridge edge images through threshold iteration based on the bridge grayscale image; according to the changes of edge pixel points in adjacent bridge edge images, the possibility that the suspected crack edges in all bridge edge images are crack edges is obtained, and several reference edges and pixels to be identified are obtained; since the pixel point with the largest gradient value in the bridge grayscale image is always located at the edge of the bridge crack, some crack edge pixel points are first obtained through the bridge edge image under the maximum threshold, and further according to the similarity of the crack edge extension direction, the crack edge pixel points in each bridge edge image are obtained by determining whether the newly added edge pixel points in the adjacent bridge edge images are similar to the crack edge pixel points in the extension direction.

[0037] Obtain the suspected splicing edge of each reference edge; obtain the splicing edge according to the shape of each reference edge and its suspected splicing edge; take all the splicing edges and the reference edge as the final crack edge in the bridge; further obtain the crack edge pixel points in the pixels to be identified according to the similarity in trend between the pixel points to be identified and the identified crack edges. Obtain the crack edge in the bridge according to the similarity between the suspected splicing edge and the reference edge. By analyzing the similarity of the crack edge extension direction under different thresholds and the coincidence of the two edges of the crack, the crack edge can be accurately identified, thereby improving the accuracy of bridge deck quality monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0039] Figure 1 It is a flow chart of the steps of the bridge deck quality monitoring method based on edge detection of the present invention;

[0040] Figure 2 This is an example of a grayscale image of a bridge;

[0041] Figure 3 This is an example diagram of the first bridge edge graph. DETAILED DESCRIPTION

[0042] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the bridge deck quality monitoring method and system based on edge detection proposed by the present invention, its specific implementation method, structure, characteristics and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0043] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0044] The specific scheme of the bridge deck quality monitoring method and system based on edge detection provided by the present invention is described in detail below with reference to the accompanying drawings.

[0045] See also Figure 1 , which shows a flow chart of the steps of a bridge deck quality monitoring method based on edge detection provided by an embodiment of the present invention, the method comprising the following steps:

[0046] Step S001: Obtain a grayscale image of a bridge.

[0047] It should be noted that cracks in bridges may affect the structural safety and bearing capacity of bridges, increase the risk of corrosion, lead to the deterioration of steel bars and concrete, and endanger traffic safety. Timely detection of cracks in bridges is necessary to formulate reasonable maintenance and reinforcement plans and extend the service life of bridges. This embodiment is a bridge deck quality monitoring method based on edge detection, specifically, accurately detecting the crack edges in bridges through machine vision; for this purpose, a grayscale image of the bridge needs to be collected first.

[0048] Specifically, a camera is used to collect the bridge surface image, the bridge surface image is grayed, and the grayed result is subjected to Gaussian filtering and noise reduction to obtain a bridge grayscale image, such as Figure 2 As shown, Figure 2 is an example of a grayscale image of a bridge; since grayscale conversion and Gaussian filtering noise reduction processing are both well-known existing technologies, they will not be described in detail in this embodiment.

[0049] Step S002: obtaining several bridge edge images through threshold iteration based on the bridge grayscale image; obtaining some crack edge pixel points according to the gradient value of the edge pixel points in the bridge edge image; obtaining the suspected crack edges in all bridge edge images according to the changes of the edge pixel points in the adjacent bridge edge images; obtaining the possibility that the suspected crack edges in all bridge edge images are crack edges according to the suspected crack edges and some crack edge pixel points in all bridge edge images; obtaining several reference edges and pixels to be identified according to the possibility that the suspected crack edges in all bridge edge images are crack edges, wherein the pixels in the reference edges are reference pixels.

[0050] It should be noted that when the camera collects the image of the bridge surface, it will be affected by the external lighting, which will cause the bridge grayscale image to have some shadow areas. The gradient value of the crack edge in the shadow area will be affected by the shadow area, resulting in the inability to detect the crack edge in the shadow position when the crack edge is monitored by the edge detection operator. In addition, due to the large grayscale difference between the shadow area and the illuminated area, the edge detection operator will identify the edge between the shadow area and the illuminated area when detecting the crack edge; that is, the crack edge in the bridge cannot be accurately obtained. Therefore, this embodiment proposes a bridge deck quality monitoring method based on edge detection.

[0051] It should be further explained that, since the pixel with the largest gradient value in the bridge grayscale image is always located at the edge of the bridge crack, the edge detection operator is used to detect the edge in the bridge grayscale image with multiple thresholds. The larger the threshold of the edge detection operator, the more edge pixels it identifies as crack edge pixels. In this way, some crack edge pixels are obtained, and the crack edge pixels are further obtained by analyzing the morphological features of the crack edge formed by the crack edge pixels and comparing them with the morphological features of the edge pixels detected under other thresholds.

[0052] Preferably, in a specific embodiment of the present invention, an initial threshold α, a termination threshold β, and a threshold iteration step γ are preset; the specific values ​​of α, β, and γ can be set by oneself in combination with actual conditions, and this embodiment does not make a rigid requirement. In this embodiment, α=0.14, β=0.04, and γ=0.02 are used as an example for description; the threshold of the edge detection operator is set to α, and the edge in the bridge grayscale image is detected using the edge detection operator to obtain a first bridge edge image, such as Figure 3 As shown, Figure 3 is an example diagram of the first bridge edge map; in this embodiment, the edge detection operator performs edge detection by using the Sobel operator, and the Sobel operator is a well-known prior art, so it will not be described in detail in this embodiment;

[0053] Further, the threshold of the edge detection operator is set to α-γ, and the edge detection operator is used to detect the edge in the bridge grayscale image to obtain a second bridge edge image;

[0054] This is repeated until the threshold of the edge detection operator is set to β, and the edge detection operator is used to detect the edge in the bridge grayscale image to obtain the xth bridge edge image, where

[0055] It should be noted that since the cracks in the bridge are slender and deep, it is difficult for external light to enter the cracks. Therefore, the gradient value of the pixel points at the edge of the crack in the bridge grayscale image is much larger than the gradient values ​​at other positions. Since the threshold of the edge detection operator for obtaining the first bridge edge image is large, most of the edge pixels in the first bridge edge image are located at the edge of the crack. The larger the gradient value, the more likely it is a crack edge pixel. Therefore, this can be used as a basis for obtaining some crack edge pixels.

[0056] Preferably, in a specific embodiment of the present invention, a crack edge gradient threshold coefficient δ is preset. The specific value of δ can be set according to actual conditions. This embodiment does not make a rigid requirement. In this embodiment, δ=0.9 is used as an example to obtain the maximum gradient value in the first bridge edge map, and the pixel points with gradient values ​​greater than δ times the maximum gradient value in the first bridge edge map are taken as crack edge pixels.

[0057] It should be noted that each bridge edge map will add a number of edge pixels compared to the previous bridge edge map, and the edge pixels include crack edge pixels and non-crack edge pixels. When a bridge is subjected to load, stress concentration will occur; and cracks usually occur along the direction of maximum stress. Since the distribution of stress in the bridge is relatively uniform, the extension directions of each position in the crack are similar. Therefore, the crack edge pixels in each bridge edge map can be obtained by analyzing the characteristics of the newly added edge pixels in each bridge edge map and whether they are similar to the crack edge pixels in the previous bridge edge map in the extension direction.

[0058] Further, for any edge pixel point in the second bridge edge map, the edge pixel point is recorded as a target point, and the pixel point with the same position coordinates as the target point in the first bridge edge map is recorded as a corresponding point of the target point. If the corresponding point is a crack edge pixel point, the target point is a crack edge pixel point. If the corresponding point is not a crack edge pixel point, the target point is a suspected crack edge pixel point. The suspected crack edge pixel points adjacent to each other in the eight neighborhoods are classified as the same suspected crack edge to obtain several suspected crack edges; the crack edge pixel points adjacent to each other in the eight neighborhoods are classified as the same crack edge to obtain several crack edges.

[0059] For any suspected crack edge, the crack edge closest to the suspected crack edge is obtained and recorded as the target edge; all pixels between the suspected crack edge and the target edge are recorded as feature points, and the possibility of the suspected crack edge being a crack edge is obtained according to the gradient direction of the pixel points corresponding to all feature points in the bridge grayscale image and the distance between the suspected crack edge and the target edge. The specific calculation formula is:

[0060] P = exp[-(σ×L)]

[0061] Wherein, P represents the possibility that the suspected crack edge is a crack edge; σ represents the standard deviation of the gradient direction of the pixel points corresponding to all feature points in the bridge grayscale image; L represents the distance between the suspected crack edge and the target edge; exp[] represents an exponential function with a natural constant as the base. This embodiment adopts the exp(-ε) model to present the inverse proportional relationship and normalization processing. ε is the input of the model. The implementer can set the inverse proportional function and normalization function according to the actual situation.

[0062] It should be noted that σ represents the standard deviation of the gradient direction of the pixel points corresponding to all feature points in the bridge grayscale image. The smaller the standard deviation, the similarity between the suspected edge and the target edge in the extension direction. Since the stress distribution in the bridge is relatively uniform, and the crack is the external manifestation of the stress distribution, the gradient values ​​between the pixels in the local range at each position in the crack are similar. The smaller the distance between the suspected edge and the target edge, the feature of similar gradient values. Therefore, the larger the value of P, the more likely the suspected crack edge is to be a crack edge.

[0063] Specifically, a possibility threshold μ of a suspected crack edge being a crack edge is preset. The specific value of μ can be set according to the actual situation. This embodiment does not make a rigid requirement. In this embodiment, μ=0.6 is used for description; if the possibility that the suspected crack edge is a crack edge is greater than or equal to the possibility threshold, the suspected crack edge is a crack edge; if the possibility that the suspected crack edge is a crack edge is less than the possibility threshold, the suspected crack edge is not a crack edge; all pixel points in the crack edge are regarded as crack edge pixel points.

[0064] Further, for any edge pixel point in the third bridge edge map, the edge pixel point is recorded as a target point, and the pixel point with the same position coordinates as the target point in the second bridge edge map is recorded as a corresponding point. If the corresponding point is a crack edge pixel point, the target point is a crack edge pixel point. If the corresponding point is not a crack edge pixel point, the target point is a suspected crack edge pixel point. The suspected crack edge pixel points adjacent to each other in the eight neighborhoods are classified as the same suspected crack edge to obtain several suspected crack edges; the crack edge pixel points adjacent to each other in the eight neighborhoods are classified as the same crack edge to obtain several crack edges.

[0065] The possibility of the suspected crack edge in the third bridge edge map being a crack edge is calculated, thereby obtaining several crack edges;

[0066] By analogy, all crack edges in the x-th bridge edge map are obtained, the crack edges are recorded as reference edges, the pixels on the reference edges are recorded as reference pixels, and all edge pixels in the x-th bridge edge map that are not reference pixels are recorded as pixels to be identified.

[0067] It should be noted that the specific process of obtaining the third bridge edge map and the crack edges in all bridge edge maps after the third bridge edge map is the same as obtaining the crack edges in all suspected crack edges in the second bridge edge map, so it will not be repeated here.

[0068] It should be further explained that the edges in the x-th bridge edge map in this step include all edges between shadow areas and illuminated areas, and all crack edges. They are obtained by analyzing the bridge edge maps under different thresholds, and the characteristics of similar extension directions of edge pixels. However, due to the influence of external light and shadow, there are unrecognized crack edges. It is also necessary to judge whether the remaining edge pixels are crack edge pixels based on the identified crack edge pixels. Therefore, all crack edge pixels identified in this step are recorded as reference pixels; the remaining edge pixels are recorded as pixels to be identified.

[0069] At this point, the reference edge, reference pixel points, and pixel points to be identified are obtained.

[0070] Step S003: According to the gradient direction of the reference pixel points in the same reference edge, obtain the standard crack width of each reference edge; according to the reference edge and its standard crack width, obtain the suspected stitching edge of each reference edge; according to the shape of the reference edge and its suspected stitching edge, obtain the similarity between each suspected stitching edge and the corresponding reference edge.

[0071] It should be noted that since the two edges of the crack in the bridge have matching characteristics, that is, the trend of one edge in the crack is highly similar to the trend of the other edge, it is possible to further obtain the crack edge pixel points in the pixel points to be identified based on the partial crack edge obtained in step S002 and the similarity in trend between the pixel points to be identified and the identified crack edges.

[0072] Preferably, in a specific embodiment of the present invention, any reference pixel point on any reference edge is recorded as a measurement point, and the distance between the reference pixel point closest to the measurement point in the gradient direction of the measurement point is recorded as the crack width of the measurement point;

[0073] Similarly, the crack widths of all reference pixel points on the reference edge line are obtained, and the average of the crack widths of all reference pixel points on the reference edge line is recorded as the standard crack width of the reference edge.

[0074] For any reference edge, a magnification coefficient τ greater than 2 is preset. The specific value of τ can be set according to the actual situation. This embodiment does not make a rigid requirement. In this embodiment, τ=3 is used for description. All pixel points to be identified whose distance from the reference edge is less than τ times the standard crack width are recorded as suspected stitching pixel points of the reference edge, and the suspected stitching pixel points adjacent to the eight neighborhoods are classified as the same suspected stitching edge, so as to obtain several suspected stitching edges of the reference edge.

[0075] It should be noted that since the cracks in the bridge have a certain width, when obtaining the crack edge pixels in the pixels to be identified based on the similarity in trend between the pixels to be identified and the identified crack edges, it is necessary to obtain the approximate width of the crack based on the identified crack edges. At the same time, a magnification factor is preset to ensure that all unidentified crack edge pixels can calculate the similarity in trend with the identified crack edge pixels.

[0076] Further, for any suspected stitching edge of any reference edge, DTW matching is performed on the reference edge and the suspected stitching edge to obtain several matching pairs. Since DTW matching is a well-known prior art, it will not be described in detail in this embodiment. According to the difference between the gradient directions of two pixels in each matching pair, the similarity between the suspected stitching edge and the reference edge is obtained. The specific calculation formula is:

[0077]

[0078] Where S represents the similarity between the suspected stitching edge and the reference edge; m represents the number of matching pairs; θ i,1,2represents the angle between the first pixel and the second pixel in the i-th matching pair in the gradient direction; cos() represents the cosine trigonometric function; || represents the absolute value function; exp[] represents an exponential function with a natural constant as the base. This embodiment adopts the exp(-ε) model to present the inverse proportional relationship and normalization processing. ε is the input of the model. The implementer can set the inverse proportional function and normalization function according to the actual situation.

[0079] It should be noted that, since the inside of the crack is dark and the outside of the crack is bright, the gradient direction of the pixel points at the edge of the crack always points vertically from the outside of the crack to the inside of the crack. Therefore, when the suspected stitched edge is a crack edge, since the suspected stitched edge and the reference edge have similar trends, the angle of the pixel points in the matching pair in the gradient direction approaches 180°. Therefore, the closer the angle of the pixel points in each matching pair in the gradient direction approaches 180°, the more similar the suspected stitched edge and the reference edge are in trend; the higher the similarity between the suspected stitched edge and the reference edge, the more likely the suspected stitched edge is the crack edge identified in step S002.

[0080] At this point, the similarity between the suspected stitching edge and the reference edge is obtained.

[0081] Step S004: obtaining a splicing edge according to the similarity between the suspected splicing edge and the corresponding reference edge; and taking all the splicing edges and the reference edges as the final crack edges in the bridge.

[0082] It should be noted that after obtaining the similarity between the suspected stitching edge and the reference edge through step S003, the similarity between the suspected stitching edge and the reference edge can be used as a basis to accurately obtain the crack edge identified in step S002; further combined with the crack edge identified in step S002, the crack edge in the bridge is accurately obtained, thereby accurately identifying the crack in the bridge.

[0083] Specifically, a similarity threshold is preset The specific value of can be set according to the actual situation. This embodiment does not make a hard requirement. For any suspected stitching edge of any reference edge, if the similarity between the suspected stitching edge and the reference edge is greater than or equal to Then the suspected joint edge is the joint edge; all the reference edges and all the joint edges in the x-th bridge edge map are taken as the final crack edges in the bridge.

[0084] At this point, this embodiment is completed.

[0085] Another embodiment of the present invention provides a bridge deck quality monitoring system based on edge detection, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the bridge deck quality monitoring method based on edge detection in steps S001 to S004 when executing the computer program.

[0086] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A bridge deck quality monitoring method based on edge detection, characterized in that: The method comprises the following steps: Get the grayscale image of the bridge; Based on the bridge grayscale image, several bridge edge images are obtained through threshold iteration; according to the gradient value of the edge pixel points in the bridge edge image, some crack edge pixel points are obtained; according to the changes of the edge pixel points in the adjacent bridge edge images, the suspected crack edges in all bridge edge images are obtained; according to the suspected crack edges and some crack edge pixel points in all bridge edge images, the possibility that the suspected crack edges in all bridge edge images are crack edges is obtained; according to the possibility that the suspected crack edges in all bridge edge images are crack edges, several reference edges and pixels to be identified are obtained, and the pixels in the reference edges are reference pixels; According to the gradient direction of the reference pixel points in the same reference edge, the standard crack width of each reference edge is obtained; according to the reference edge and its standard crack width, the suspected splicing edge of each reference edge is obtained; according to the shapes of the reference edge and its suspected splicing edge, the similarity between each suspected splicing edge and the corresponding reference edge is obtained; According to the similarity between the suspected joint edge and the corresponding reference edge, the joint edge is obtained; all the joint edges and the reference edges are taken as the final crack edges in the bridge.

2. The bridge deck quality monitoring method based on edge detection according to claim 1 is characterized in that: The specific method of obtaining a plurality of bridge edge images through threshold iteration based on the bridge grayscale image is as follows: Preset an initial threshold α, a termination threshold β, and a threshold iteration step length γ; set the threshold of the edge detection operator to α, use the edge detection operator to detect the edge in the bridge grayscale image, and obtain a first bridge edge image; The threshold of the edge detection operator is set to α-γ, and the edge detection operator is used to detect the edge in the bridge grayscale image to obtain a second bridge edge image; Until the threshold of the edge detection operator is set to β, the edge detection operator is used to detect the edge in the bridge grayscale image to obtain the xth bridge edge image, where 3. The bridge deck quality monitoring method based on edge detection according to claim 1 is characterized in that: The specific method of obtaining some crack edge pixel points according to the gradient value of the edge pixel points in the bridge edge map includes: A crack edge gradient threshold coefficient δ is preset, the maximum gradient value in the first bridge edge map is obtained, and the pixel points in the first bridge edge map whose gradient values ​​are greater than δ times the maximum gradient value are taken as crack edge pixels.

4. The bridge deck quality monitoring method based on edge detection according to claim 1 is characterized in that: The specific method of obtaining the suspected crack edges in all bridge edge maps according to the changes of edge pixels in adjacent bridge edge maps is as follows: For any edge pixel point in the second bridge edge map, the edge pixel point is recorded as the target point, and the pixel point in the first bridge edge map with the same position coordinates as the target point is recorded as the corresponding point of the target point. If the corresponding point is a crack edge pixel point, the target point is the crack edge pixel point. If the corresponding point is not a crack edge pixel point, the target point is a suspected crack edge pixel point. The adjacent suspected crack edge pixel points in the eight neighborhoods are classified as the same suspected crack edge to obtain several suspected crack edges; the adjacent crack edge pixel points in the eight neighborhoods are classified as the same crack edge to obtain several crack edges.

5. The bridge deck quality monitoring method based on edge detection according to claim 4 is characterized in that: The method of obtaining the possibility that the suspected crack edges in all bridge edge maps are crack edges according to the suspected crack edges and partial crack edge pixels in all bridge edge maps includes: For any suspected crack edge, the crack edge closest to the suspected crack edge is obtained and recorded as the target edge; all pixels between the suspected crack edge and the target edge are recorded as feature points, and the possibility of the suspected crack edge being a crack edge is obtained according to the gradient direction of the pixel points corresponding to all feature points in the bridge grayscale image and the distance between the suspected crack edge and the target edge. The specific calculation formula is: P = exp[-(σ×L)] Wherein, P represents the possibility that the suspected crack edge is a crack edge; σ represents the standard deviation of the gradient direction of the pixel points corresponding to all feature points in the bridge grayscale image; L represents the distance between the suspected crack edge and the target edge; exp[] represents an exponential function with a natural constant as the base.

6. The bridge deck quality monitoring method based on edge detection according to claim 5 is characterized in that: The method of obtaining a plurality of reference edges and pixels to be identified according to the possibility that the suspected crack edges in all bridge edge images are crack edges includes: A possibility threshold μ of the suspected crack edge being a crack edge is preset. If the possibility that the suspected crack edge is a crack edge is greater than or equal to the possibility threshold, the suspected crack edge is a crack edge; if the possibility that the suspected crack edge is a crack edge is less than the possibility threshold, the suspected crack edge is not a crack edge; all pixel points in the crack edge are regarded as crack edge pixel points; For any edge pixel point in the third bridge edge map, the edge pixel point is recorded as a target point, and the pixel point with the same position coordinates as the target point in the second bridge edge map is recorded as a corresponding point. If the corresponding point is a crack edge pixel point, the target point is a crack edge pixel point. If the corresponding point is not a crack edge pixel point, the target point is a suspected crack edge pixel point. The adjacent suspected crack edge pixel points in the eight neighborhoods are classified as the same suspected crack edge to obtain several suspected crack edges; the adjacent crack edge pixel points in the eight neighborhoods are classified as the same crack edge to obtain several crack edges. The possibility of the suspected crack edge in the third bridge edge map being a crack edge is calculated, thereby obtaining several crack edges; By analogy, all crack edges in the x-th bridge edge map are obtained, the crack edges are recorded as reference edges, the pixels on the reference edges are recorded as reference pixels, and all edge pixels in the x-th bridge edge map that are not reference pixels are recorded as pixels to be identified.

7. The bridge deck quality monitoring method based on edge detection according to claim 1 is characterized in that: The specific method of obtaining the suspected splicing edge of each reference edge according to the reference edge and its standard crack width includes: For any reference edge, a magnification coefficient τ is preset; all pixel points to be identified whose distance from the reference edge is less than τ times its standard crack width are recorded as suspected stitching pixel points of the reference edge, and the suspected stitching pixel points adjacent to the eight neighborhoods are classified as the same suspected stitching edge, so as to obtain several suspected stitching edges of the reference edge.

8. The bridge deck quality monitoring method based on edge detection according to claim 1 is characterized in that: The specific method of obtaining the similarity between each suspected splicing edge and the corresponding reference edge according to the shapes of the reference edge and the suspected splicing edge is as follows: For any suspected stitching edge of any reference edge; perform DTW matching on the reference edge and the suspected stitching edge to obtain several matching pairs; obtain the similarity between the suspected stitching edge and the reference edge according to the difference between the gradient directions of two pixels in each matching pair, and the specific calculation formula is: Where S represents the similarity between the suspected stitching edge and the reference edge; m represents the number of matching pairs; θ i,1,2 It represents the angle between the first pixel and the second pixel in the i-th matching pair in the gradient direction; cos() represents the cosine trigonometric function; || represents the absolute value function; exp[] represents the exponential function with a natural constant as the base.

9. The bridge deck quality monitoring method based on edge detection according to claim 1 is characterized in that: The specific method of obtaining the stitching edge according to the similarity between the suspected stitching edge and the corresponding reference edge is as follows: Preset a similarity threshold For any suspected stitching edge of any reference edge, if the similarity between the suspected stitching edge and the reference edge is greater than or equal to Then the suspected stitching edge is the stitching edge.

10. A bridge deck quality monitoring system based on edge detection, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the computer program is executed by a processor, the steps of the bridge deck quality monitoring method based on edge detection as described in any one of claims 1 to 9 are implemented.

Images