CN117974753A - Bridge tunnel crack depth measurement method - Google Patents
Bridge tunnel crack depth measurement method Download PDFInfo
- Publication number
- CN117974753A CN117974753A CN202410384276.8A CN202410384276A CN117974753A CN 117974753 A CN117974753 A CN 117974753A CN 202410384276 A CN202410384276 A CN 202410384276A CN 117974753 A CN117974753 A CN 117974753A
- Authority
- CN
- China
- Prior art keywords
- crack
- matching
- region
- matching window
- branch
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000691 measurement method Methods 0.000 title claims abstract description 7
- 238000012216 screening Methods 0.000 claims abstract description 8
- 238000000034 method Methods 0.000 claims description 31
- 238000004364 calculation method Methods 0.000 claims description 19
- 230000000877 morphologic effect Effects 0.000 claims description 12
- 239000011159 matrix material Substances 0.000 claims description 11
- 238000004458 analytical method Methods 0.000 claims description 9
- 238000010606 normalization Methods 0.000 claims description 6
- 238000007670 refining Methods 0.000 claims description 5
- 238000001514 detection method Methods 0.000 abstract description 15
- 238000005286 illumination Methods 0.000 abstract description 13
- 230000000694 effects Effects 0.000 abstract description 10
- 230000007547 defect Effects 0.000 abstract description 5
- 238000005259 measurement Methods 0.000 abstract description 5
- 238000006243 chemical reaction Methods 0.000 description 6
- 230000009466 transformation Effects 0.000 description 5
- 230000008859 change Effects 0.000 description 4
- 238000005336 cracking Methods 0.000 description 3
- 238000001914 filtration Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000003044 adaptive effect Effects 0.000 description 2
- 230000002776 aggregation Effects 0.000 description 2
- 238000004220 aggregation Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 229910000831 Steel Inorganic materials 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002146 bilateral effect Effects 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000007797 corrosion Effects 0.000 description 1
- 238000005260 corrosion Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000003708 edge detection Methods 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 239000010959 steel Substances 0.000 description 1
Landscapes
- Image Analysis (AREA)
Abstract
The invention relates to the technical field of crack depth measurement, in particular to a bridge tunnel crack depth measurement method. Firstly, obtaining binocular images of a bridge tunnel; sliding a matching window on any view in the binocular image, continuously changing parallax to obtain a corresponding window corresponding to the matching window in the other view, and determining a matching cost weight coefficient of each matching window; weighting matching windows in the binocular image and matching costs corresponding to the corresponding windows based on the matching cost weight coefficient, updating the matching costs, and screening out optimal parallax and a corresponding parallax map according to the continuously updated matching costs; and determining the crack depth of the bridge tunnel based on the disparity map. According to the invention, the crack depth is measured through the stereo matching algorithm, so that the potential risk of the bridge can be estimated, the defect of low precision under insufficient texture and illumination conditions is overcome, the reliability of depth detection is improved, and the effect of detecting the crack depth is improved.
Description
Technical Field
The invention relates to the technical field of crack depth measurement, in particular to a bridge tunnel crack depth measurement method.
Background
Bridges are a vital part of traffic infrastructure and play an important role in the smoothness of traffic and the safety of people's lives and properties. In the daily use process, the bridge may have cracks due to factors such as load bearing, temperature change, corrosion of the steel bars and the like. If the cracks are not found and treated in time, the functions of the bridge are affected, the service life of the bridge is shortened, when the functions of the bridge are affected, the bridge is light and is difficult to resist the influence of external natural environment, and the bridge deck is broken, so that people flow and vehicles on the bridge are dropped.
The conventional method for detecting the depth of the crack of the bridge tunnel comprises the steps of obtaining a binocular image of the bridge tunnel, and directly detecting the depth of the crack of the obtained binocular image, wherein the problem of inaccurate crack depth detection caused by the fact that the binocular image is easily interfered by a complex background exists. These interferents may make detection of crack depths more difficult due to the large amount of complex background information that may exist around the bridge. In addition, the binocular image is also easily affected by illumination and texture, so that the obtained binocular image is directly subjected to crack depth detection, and the conditions of poor detection effect, and accuracy and stability to be improved exist.
Disclosure of Invention
In order to solve the technical problems that the detection effect is poor, the accuracy and the stability are to be improved when the obtained binocular image is directly subjected to crack depth detection, the invention aims to provide a bridge tunnel crack depth measurement method, and the adopted technical scheme is as follows:
Obtaining binocular images of a bridge tunnel; acquiring a crack region from the binocular image;
sliding a matching window on any view in the binocular image, continuously changing parallax to obtain at least two corresponding windows corresponding to the matching window in the other view, and determining a crack texture significant coefficient of the matching window according to the image gray scale characteristics of the corresponding region of the matching window and the distance between the center of the matching window and the crack region;
Carrying out morphological refinement operation and branch point analysis on the crack region, and constructing crack branch point extensibility of a branch point on the edge line of the crack region; identifying corner points at the intersections of the cracks in the matching window, and analyzing by combining the corner points and the extensibility of the branching points of the cracks to determine the complexity coefficients of the branching of the cracks of the matching window;
Combining the crack texture significant coefficient and the crack branch complex coefficient to determine a matching cost weight coefficient of each matching window; and weighting matching costs corresponding to the matching windows and corresponding windows in the binocular image based on the matching cost weight coefficient, updating the matching costs, screening out optimal parallax and a corresponding parallax map according to the continuously updated matching costs, and determining the crack depth of the bridge tunnel based on the parallax map.
Preferably, after the obtaining the crack region from the binocular image, the method further includes: and analyzing the gray level of the crack region to determine a deep crack region and a shallow crack region.
Preferably, the determining the crack texture saliency coefficient of the matching window according to the image gray scale characteristics of the corresponding area of the matching window and the distance between the center of the matching window and the crack area includes:
Calculating the inverse variance and contrast of the gray level co-occurrence matrix of the corresponding region of the matching window; and determining the crack texture significant coefficient of the matching window by combining the inverse variance, the contrast, the number of pixels in the deep crack region, the number of pixels in the shallow crack region and the distances between the center of the matching window and the edge lines of the deep crack region and the shallow crack region.
Preferably, the calculation formula of the crack texture significant coefficient is as follows:;
Soct is the crack texture significant coefficient of the matching window; CON is the contrast of the gray level co-occurrence matrix of the corresponding region of the matching window; h is the inverse variance of the gray level co-occurrence matrix of the region corresponding to the matching window; nd is the number of pixel points in the deep crack area; ns is the number of pixel points in the shallow crack region; ds is the shortest distance between the center of the matching window and the edge line of the nearest shallow crack region; dd is the shortest distance between the center of the matching window and the edge line of the nearest deep crack region; norm is a normalization function.
Preferably, the determining the deep crack region and the shallow crack region includes:
Clustering crack areas based on gray values to obtain two categories;
calculating the gray average value of the crack area in the two categories;
Taking a crack region in the class with the smallest gray average value in the two classes as a deep crack region; and taking a crack region in the category with the maximum gray average value in the two categories as a shallow crack region.
Preferably, the performing morphological refinement operation and branch point analysis on the crack region to construct a crack branch point extensibility of a branch point on an edge line of the crack region includes:
carrying out morphological refining operation on the crack region, and refining the crack region into crack lines;
Taking a crack line corresponding to a crack region with the largest width as a first-stage crack; acquiring branch points of other cracks on the first-stage crack, marking the branch points as first-stage branch points, and taking crack lines extending from the first-stage branch points as second-stage cracks; acquiring branch points of other cracks on the secondary cracks, marking the branch points as secondary branch points, and taking crack lines extending from the secondary branch points as tertiary cracks; and so on until there are no branching points and cracks;
And for the crack lines corresponding to any crack region, determining the crack branch point extensibility of the branch point by combining the included angle between the crack lines and the crack lines extending at each branch point on the crack lines and the number of cracks at the same branch point, wherein the included angle between the crack lines extending at each branch point on the crack lines and the number of cracks at the same branch point are in positive correlation with the crack branch point extensibility.
Preferably, the analyzing the joint angular point and the crack branching point extensibility to determine the crack branching complexity coefficient of the matching window includes:
clustering the corner points in the matching window to obtain a cluster;
when the corner points are branch points, taking the extension degree of the crack branch points corresponding to the corner points as a molecule, taking the sum value of Euclidean distances between two corner points in a cluster to which the corner points belong as a denominator, obtaining a first ratio, and taking the first ratio as the branch point density degree of the corner points;
and taking the sum of the branch point densities corresponding to all the corner points as a crack branch complexity coefficient of the matching window.
Preferably, the determining the matching cost weight coefficient of each matching window by combining the crack texture significant coefficient and the crack branch complex coefficient includes:
Taking the preset parameters as molecules, taking the sum of the crack texture significant coefficient of each matching window and the crack branch complex coefficient of each matching window as denominators to obtain a second ratio, and carrying out inverse proportion normalization on the second ratio to obtain a matching cost weight coefficient of each matching window; wherein the preset parameter is a positive number.
Preferably, the calculation formula of the update matching cost is as follows:
;
wherein C (p, q) is the updated matching cost; Matching the cost weight coefficient; ((p,q), ) Is a Census-based cost; ((p,q), ) At the cost of AD.
Preferably, the acquiring the crack region from the binocular image includes:
crack regions in the binocular image are identified by Niblack thresholding.
The embodiment of the invention has at least the following beneficial effects:
The invention relates to the technical field of crack depth measurement. Firstly, acquiring a crack region from a binocular image of a bridge tunnel so as to facilitate subsequent analysis of crack textures, and acquiring a matching cost weight coefficient through the crack textures; sliding a matching window on any view in the binocular image, continuously changing parallax to obtain at least two corresponding windows corresponding to the matching window in the other view, determining a crack texture significant coefficient of the matching window, wherein the crack texture significant coefficient reflects the condition of rich textures of a crack region, and correspondingly adjusting the weight of matching cost when the textures are richer; the crack is easy to generate tiny branches, and the branch characteristics are that the branches extend from the main crack area to all directions, so that morphological refinement operation and branch point analysis are further carried out on the crack area, and the branch point extensibility of the crack is determined; because the surface of the crack branch is uneven and the ravines are obvious, the texture of the crack branch affects the cost calculation in the matching window, the corner points at the crack intersection in the matching window are identified, and the corner points and the crack branch point extensibility are combined for analysis to determine the crack branch complexity coefficient of the matching window; combining the crack texture significant coefficient and the crack branch complex coefficient to determine a matching cost weight coefficient of each matching window; and weighting matching costs corresponding to the matching windows and corresponding windows in the binocular image based on the matching cost weight coefficient, updating the matching costs, screening out optimal parallax and a corresponding parallax map according to the continuously updated matching costs, and determining the crack depth of the bridge tunnel based on the parallax map. According to the method, the potential risk of the bridge can be estimated by measuring the cracking depth through the stereo matching algorithm according to the binocular image of the bridge tunnel shot by the binocular stereo camera, the defect that the precision of the traditional detection method is low under the condition of insufficient texture and illumination is overcome, the reliability of depth detection is improved, and the effect of detecting the crack depth is improved.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flow chart of a method for measuring a crack depth of a bridge tunnel according to an embodiment of the present invention.
Detailed Description
In order to further describe the technical means and effects adopted by the invention to achieve the preset aim, the following detailed description is given below of a bridge tunnel crack depth measuring method according to the invention, which is specific to the implementation, structure, characteristics and effects thereof, with reference to the accompanying drawings and the preferred embodiment. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
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.
The embodiment of the invention provides a concrete implementation method of a bridge tunnel crack depth measurement method, which is suitable for bridge building design scenes. And shooting the bridge cracks by using a binocular stereo camera to obtain binocular images of two views. In order to solve the technical problems that the detection effect is poor, and the accuracy and stability are to be improved when the crack depth detection is directly carried out on the obtained binocular image. According to the method, the potential risk of the bridge can be estimated by measuring the cracking depth through the stereo matching algorithm according to the binocular image of the bridge tunnel shot by the binocular stereo camera, the defect that the precision of the traditional detection method is low under the condition of insufficient texture and illumination is overcome, the reliability of depth detection is improved, and the effect of detecting the crack depth is improved.
The invention provides a concrete scheme of a bridge tunnel crack depth measuring method by combining a drawing.
Referring to fig. 1, a flowchart of a method for measuring a depth of a crack in a bridge tunnel according to an embodiment of the invention is shown, and the method includes the following steps:
step S100, obtaining binocular images of a bridge tunnel; the crack region is acquired from the binocular image.
The bridge crack is shot by using the binocular stereo camera, an initial bridge crack image of two views is obtained, and the shot image is dark due to insufficient illumination at the crack, so that key information is not easy to obtain, and the stereo matching precision is affected to a certain extent. Therefore, the initial bridge crack image is preprocessed, and Gamma transformation is adopted to enhance the image, so that the identifiability is improved, and the contrast of the image is increased. And then filtering noise in the image by adopting a median filtering method to obtain a binocular image containing two views.
Stereo matching is a technique for recovering depth information of a real scene from a planar image by finding matching point pairs from two or more images of the same scene and then calculating the depth of spatial physical points corresponding to the point pairs according to the principle of triangulation.
The AD-Census algorithm is an algorithm combining a local algorithm and a semi-global algorithm, has robustness and high accuracy, performs matching cost calculation in the first step of three-dimensional matching, is a measurement parameter of similarity, reflects gray level change conditions of pixel points based on matching cost conversion (Adaptive Census Transform, AD) calculated by single pixel points, and can achieve good matching effect in areas with rich textures. However, the illumination variation may have a large influence on the AD conversion, resulting in relatively low accuracy of image matching in the case of uneven illumination. In contrast, census transformation has good adaptability, can effectively cope with illumination variation, and provides good correction effect when gray level variation caused by uneven illumination exists, and has the defect that a characteristic descriptor is too single, and when gray level variation fluctuation in a neighborhood window is large or repeated texture areas appear, matching cost obtained by Census transformation is difficult to well reflect similarity. The weights can be distributed for AD conversion and Census conversion in a self-adaptive mode according to the texture characteristics of the binocular image of the bridge tunnel.
Taking AD as an example, the matching cost is calculated according to the following formula:
(p,q)=;
wherein the point to be matched in the left image in the binocular vision image is p (x, y), and the gray scale is ; The pixel point with the parallax d in the transverse direction of the right image is q ((x-d), y), and the gray scale is((x-d),y);(P, q) is the matching cost of AD. In the invention, the parallax d is continuously changed so as to find the optimal parallax later and obtain a corresponding parallax map. The method comprises the steps of calculating hamming distances of matching windows in left and right views through Census transformation, wherein gray values in the centers of the matching windows are used as standard gray values, when gray values of pixel points in the matching windows are larger than the standard gray values and marked as 0, gray values of the pixel points in the matching windows are smaller than or equal to the standard gray values and marked as 1, performing exclusive OR operation on the obtained two binary codes, counting the number of 1, and indicating that the similarity of the two matching windows is higher when the number is smaller.
The conventional calculation formula of the AD-Census matching cost is defined as:
;
Wherein C (p, q)' is a conventional matching cost; ((p,q), ) Is a Census-based cost; ((p,q), ) Is based on the cost of AD; For varying influencing parameters, C and All take positive values, can passThe exponential transformation of the function normalizes it to within (0, 1).
In the invention, the matching cost corresponding to the matching windows in the two views of the binocular image is weighted, so that the matching cost is adjusted by utilizing the texture and gray information of the two views.
The updated matching cost calculation formula is as follows: c (p, q) = (1-)((p,q),)+();
Wherein C (p, q) is the updated matching cost; Matching the cost weight coefficient; is a Census-based cost; At the cost of AD.
If a certain area of the image is susceptible to illumination changes, the weight of Census changes should be increased, i.eIf the texture of a certain region is more, the AD change weight needs to be increased to match the cost weight coefficientThe larger. To determine whether the window of matching costs is in a region of complex texture or a region of large illumination variation, it can be determined by analyzing texture features. The crack area can be divided into a deep crack and a shallow crack, the deep crack is greatly influenced by illumination, the color is deep, texture information is hardly seen, and the shallow crack can reflect part of light, but the surface is damaged, and the crack is in a rough and fine stripe pattern. And designing adaptive matching cost coefficients for different areas.
Because the crack area has larger phase difference with the background, the color of the crack area is darker, the background wall part is brighter, and the obvious difference exists. Further, acquiring a crack region from the binocular image includes: crack regions in the binocular image are identified by Niblack thresholding. The binocular image is binarized by using Niblack threshold algorithm, so that the whole crack region in the binocular image can be directly segmented, and the region with low gray value is marked as the crack region. And (3) for the binocular image of the split crack region, using a canny edge detection operator to obtain the edge of the crack region, marking the edge as a crack region edge line, and obtaining a binary image as a result, wherein the crack region edge line comprises the outer boundary of the shallow crack region and the inner boundary of the deep crack region.
Step S200, sliding a matching window on any view in the binocular image, continuously changing parallax, obtaining at least two corresponding windows corresponding to the matching window in another view, and determining a crack texture significant coefficient of the matching window according to the image gray scale characteristics of the corresponding region of the matching window and the distance between the center of the matching window and the crack region.
Different textures exist in a fixed window selected by the stereo matching algorithm, and the weight coefficient is changed according to the definition of the textures. The window comprises shallow crack parts with clear textures and deep crack parts with fuzzy textures in terms of the area.
After the crack region is obtained from the binocular image, the gray scale of the crack region is analyzed, and the deep crack region and the shallow crack region are determined. Therefore, the two types of areas are divided by using a fuzzy c-means clustering algorithm, the clustering number is set to be 2, the deep cracks mainly show black, and the gray value of the pixel points is smaller than that of the pixel points of the shallow cracks. And inputting the gray value of the pixel point of the crack region, and outputting the gray value as the average value of the gray values of the two categories. One of the two classes with the smallest gray average value is marked as a deep crack, and the other class is marked as a shallow crack. That is, based on the gray value, clustering the crack areas to obtain two categories; calculating the gray average value of the crack area in the two categories; taking a crack region in the class with the smallest gray average value in the two classes as a deep crack region; and taking a crack region in the category with the maximum gray average value in the two categories as a shallow crack region. The number of pixels in the shallow crack area is recorded as Ns, and the number of pixels in the deep crack area is recorded as Nd.
And sliding the matching window on any view in the binocular image, and continuously changing parallax to obtain at least two corresponding windows corresponding to the matching window in the other view. In the embodiment of the invention, the matching windows are slid on the left view in the binocular image, under the condition of continuously changing the parallax, corresponding windows corresponding to the matching windows are continuously obtained on the right view, namely, for each matching window, each parallax is changed once, a corresponding window is arranged on the right view, and each matching window on the left view is provided with a plurality of corresponding windows in the right view. The center point of the matching window on the left view is changed continuously, that is, the matching window is slid continuously, and a plurality of corresponding windows are corresponding to each sliding matching window synchronously.
And calculating a gray level co-occurrence matrix corresponding to the region in the matching window, and calculating two characteristic values of the inverse variance H and the contrast CON. The inverse variance reflects the degree of variation of the texture and the contrast reflects the depth of the image texture ravines. As the texture of the boundary is more obvious than that of the inside of the region, the closer the pixels are to the boundary, the larger the influence of the texture on the matching cost calculation is, the shortest distance between the center pixel point of the matching window and the edge line of the nearest shallow crack region is calculated as Ds, and the shortest distance between the center pixel point of the matching window and the edge line of the nearest deep crack region is calculated as Dd. And constructing a crack texture saliency coefficient soct according to the related indexes. According to the image gray scale characteristics of the corresponding area of the matching window and the distance between the center of the matching window and the crack area, the crack texture significant coefficient of the matching window is determined, and the specific is that:
Calculating the inverse variance and contrast of the gray level co-occurrence matrix of the corresponding region of the matching window; and determining the crack texture significant coefficient of the matching window by combining the inverse variance, the contrast, the number of pixels in the deep crack region, the number of pixels in the shallow crack region and the distances between the center of the matching window and the edge lines of the deep crack region and the shallow crack region.
The calculation formula of the crack texture significant coefficient is as follows:;
Soct is the crack texture significant coefficient of the matching window; CON is the contrast of the gray level co-occurrence matrix of the corresponding region of the matching window; h is the inverse variance of the gray level co-occurrence matrix of the region corresponding to the matching window; nd is the number of pixel points in the deep crack area; ns is the number of pixel points in the shallow crack region; ds is the shortest distance between the center of the matching window and the edge line of the nearest shallow crack region; dd is the shortest distance between the center of the matching window and the edge line of the nearest deep crack region; norm is a normalization function.
The larger the value of the inverse variance H, the higher the degree of variation of the texture; the contrast CON becomes larger and the texture grooves become deeper. And the higher the pixel number of the shallow cracks is, the larger the pixel number Ns in the shallow crack area is, and the smaller the pixel number Nd in the deep crack area is, the smaller the Nd/Ns ratio is, which means that the more the texture detail part in the matching window area is. The smaller the values of the shortest distance Ds between the center of the matching window and the edge line of the nearest shallow crack region and the shortest distance Dd between the center of the matching window and the edge line of the nearest deep crack region, the closer the boundary distance between the center of the window and the deep and shallow crack is, the more likely the boundary line passes through the inside of the window, and the weight of the matching cost is changed under the influence of boundary textures. The larger the numerator is, the larger the value of soct is, the larger the significant coefficient of the crack texture is, which means that the weight of AD conversion is increased in the region with rich textures.
Step S300, performing morphological refinement operation and branch point analysis on the crack region, and constructing crack branch point extensibility of a branch point on an edge line of the crack region; and identifying corner points at the intersections of the cracks in the matching window, and analyzing by combining the corner points and the extensibility of the branching points of the cracks to determine the complexity coefficients of the branching of the cracks of the matching window.
The deep cracks are relatively regular in morphology, while shallow cracks are prone to fine branches, which are characterized by stretching from the main crack region in all directions. The method is characterized in that tiny crack branches appear on a main crack, other branches exist on the crack branches, and the cracks are classified according to the width of the cracks.
In the digital image, the crack has width, length and direction, the width value of each position of the crack is calculated, a plurality of points are randomly selected in the crack to calculate the maximum inscribed circle of the boundary, and the diameter of the inscribed circle is the width of the crack. And carrying out morphological thinning operation on the crack region to refine the crack into lines with single pixel width. Firstly, taking a crack with the largest width as a first-stage crack, searching a branch point according to pixels from one end point of the first-stage crack, checking whether a line corresponding to the first-stage crack has branches in other directions in an image after morphological refinement, and if more than two adjacent pixel points exist in 8 adjacent pixel points, indicating that the branches exist, and marking the branches as the first-stage branch point. All cracks extending on the first-stage branch point are marked as second-stage cracks, whether the branch point exists in the second-stage cracks is checked, if yes, the second-stage branch point is marked, and the like. Then find the branching point of the next stage and the crack of the next stage on the crack according to the method.
Specifically, morphological refining operation is carried out on the crack region, and the crack region is refined into crack lines; taking a crack line corresponding to a crack region with the largest width as a first-stage crack; acquiring branch points of other cracks on the first-stage crack, marking the branch points as first-stage branch points, and taking crack lines extending from the first-stage branch points as second-stage cracks; acquiring branch points of other cracks on the secondary cracks, marking the branch points as secondary branch points, and taking crack lines extending from the secondary branch points as tertiary cracks; and so on until there are no branching points and cracks;
And for the crack lines corresponding to any crack region, determining the crack branch point extensibility of the branch point by combining the included angle between the crack lines and the crack lines extending at each branch point on the crack lines and the number of cracks at the same branch point, wherein the included angle between the crack lines extending at each branch point on the crack lines and the number of cracks at the same branch point are in positive correlation with the crack branch point extensibility.
The method for acquiring the included angle between the crack lines and the crack lines extending at each branching point on the crack lines comprises the following steps: and for the crack line a, acquiring a pixel point closest to the crack line a in a crack line c extending at a branching point b on the crack line a, connecting the previous pixel point of the branching point b with the branching point along the direction of the crack line a to obtain a line segment e, connecting the branching point with the extending point d to obtain a line segment f, taking an included angle formed by the line segment e and the line segment f as an included angle between the crack line a and the crack line c, and normalizing the included angle to eliminate the influence of dimension.
As an embodiment of the present invention, the calculation formula of the extensibility of the branching point of the crack is:;
Wherein, Crack branch point extensibility as the i-th branch point; Is the number of cracks at the ith branch point; e is a natural constant, and angle is the average value of the included angle between the crack line and the crack line extending at the ith branching point of the crack line. The number of the crack lines extending at the branching point is at least one.
The number Brank of cracks at the branching point becomes larger, which indicates that the more the branching point cracks have, the more the texture details of the local area are, the larger the average angle of the included angles is, and the more obvious the details among the image cracks are. The number Brank of cracks at the branching point and the average angle of the included angles become larger, the crack branching point spread degree ed becomes larger, the crack branching point spread degree reflects the detail of the crack branching point, and the larger the crack branching point spread degree is, the weight of AD change needs to be increased in cost calculation is reflected.
The surface of the branch is uneven, the ravines are obvious, and the texture of the branches can influence the cost calculation in the matching window. Identifying corner points at crack intersections in the matching window, and particularly: and (3) detecting and identifying the corner points of each matching window in the binocular image through the harris corner points, wherein the corner points are easy to densely appear at the intersection of the crack branches and the main crack. Density-based clustering algorithm (Density-Based Spatial Clustering of Applications with Noise, DBSCAN) can be used for Density clustering of the corner points, the spatial position coordinates of each corner point are input, and the clustering clusters of the corner points are output. Further, the crack branch complex coefficients of the matching window are determined by analyzing the corner points and the crack branch point extensibility, and the crack branch complex coefficients are specifically:
Clustering the corner points in the matching window to obtain a cluster; when the corner points are branch points, taking the extension degree of the crack branch points corresponding to the corner points as a molecule, taking the sum value of Euclidean distances between two corner points in a cluster to which the corner points belong as a denominator, obtaining a first ratio, and taking the first ratio as the branch point density degree of the corner points; and taking the sum of the branch point densities corresponding to all the corner points as a crack branch complexity coefficient of the matching window. When the corner point is not a branching point, the branching point density of the corner point is recorded as 0.
Forming corner pairs by two corner points belonging to the same cluster;
when the corner point is a branching point, the calculation formula of the crack branching complex coefficient is as follows: ;
Fbci is the crack branching complexity coefficient of the matching window; numf is the total number of corner points within the matching window that are branch points; Crack branching point extensibility which is the j-th corner point of the branching point, Is the sum of Euclidean distances between two corner points in the cluster of the corner points of the j-th branch point.
The greater the number of corner points in the matching window, the smaller the euclidean distance between two corner points,The larger the value of (c) is, the denser the branch points in the matching window are; meanwhile, when the extension degree of the crack branch points of the branch points in the matching window is larger, the number of the extended crack lines of the corresponding branch points is larger, and the complexity of the crack branches in the corresponding matching window is larger. Conversely, the fewer the number of corner points in the matching window, the greater the Euclidean distance between two corner points, and the greater the distance between the two corner pointsThe smaller the value of (2) is, the more scattered the branch points in the matching window are, and the smaller the corresponding crack branch complex coefficient is; meanwhile, when the extension degree of the crack branch points of the branch points in the matching window is smaller, the number of the extended crack lines of the corresponding branch points is smaller, and the complexity of the crack branches in the corresponding matching window is smaller.
Step S400, combining the crack texture significant coefficient and the crack branch complex coefficient to determine a matching cost weight coefficient of each matching window; and weighting matching costs corresponding to the matching windows and corresponding windows in the binocular image based on the matching cost weight coefficient, updating the matching costs, screening out optimal parallax and a corresponding parallax map according to the continuously updated matching costs, and determining the crack depth of the bridge tunnel based on the parallax map.
Combining the crack texture significant coefficient and the crack branch complex coefficient, determining a matching cost weight coefficient of each matching window, and specifically: taking the preset parameters as molecules, taking the sum of the crack texture significant coefficient of each matching window and the crack branch complex coefficient of each matching window as denominators to obtain a second ratio, and carrying out inverse proportion normalization on the second ratio to obtain a matching cost weight coefficient of each matching window; wherein the preset parameter is a positive number. In the embodiment of the invention, the preset parameter has a value of 0.5, and in other embodiments, the value is adjusted by an implementer according to the actual situation.
The calculation formula of the matching cost weight coefficient is as follows:;
Wherein, Matching cost weight coefficients for the matching windows; e is a natural constant; is a preset parameter; fbci is the crack branching complexity coefficient of the matching window; soct is the crack texture saliency coefficient of the matching window.
An exponential function with a natural constant e as a base number is used for realizing the matchingWhen the crack texture significant coefficient soct and the crack branch complex coefficient fbci are higher, the region with clear texture and more details is indicated to be matched with the cost weight coefficientThe larger the value of (a) is, the higher the weight of the AD conversion is. And calculating the cost by using different weights to obtain a similarity result of the left image and the right image.
Further, after determining the matching cost weight coefficient of each matching window, weighting the matching cost corresponding to the matching window in the two views of the binocular image based on the matching cost weight coefficient, updating the matching cost, and obtaining the parallax image based on the updated matching cost.
The calculation formula for updating the matching cost is already given in step S200, and the specific matching cost weight coefficient is substituted into the calculation formula for updating the matching cost, so that the updating of the matching cost can be realized.
And screening out the optimal parallax and the corresponding parallax map according to the continuously updated matching cost. It should be noted that, based on the continuously updated matching cost, the method of screening out the optimal parallax and the corresponding parallax map is a well-known technique of those skilled in the art, which is a step in the AD-Census algorithm, and specific details thereof will not be repeated.
Next, carrying out cost aggregation, and adopting a path aggregation method in a stereo matching algorithm (SGM), wherein the step is used for establishing connection of adjacent pixel values and optimizing a cost matrix. And then determining an optimal parallax value of each pixel through parallax calculation, and selecting the parallax corresponding to the minimum cost value as the optimal parallax from cost values of all parallaxes of a certain pixel. And finally, performing parallax optimization, namely removing error parallax caused by noise and shielding by using a common left-right consistency check algorithm in stereo matching, and smoothing a parallax map by using bilateral filtering to obtain the parallax map.
After the disparity map is obtained, determining the crack depth of the bridge tunnel based on the disparity map, namely further obtaining the crack depth of the bridge tunnel according to a disparity result, wherein the calculation formula is as follows:;
z is the crack depth of the bridge tunnel; b is the optical center distance of the binocular camera, i.e. the distance between the two viewing angles; f is the focal length of the binocular camera; d is the optimal parallax for the two views in the binocular image. It should be noted that, in the actual operation process, the optical center distance b of the binocular camera and the focal length f of the binocular camera are set in advance. And calculating parallax to obtain a depth map, wherein the parallax and the depth map are inversely proportional. The crack depth is the difference between the center depth value of the crack and the surrounding depth values of the crack.
In summary, the invention relates to the technical field of crack depth measurement. Firstly, obtaining binocular images of a bridge tunnel; acquiring a crack region from the binocular image; sliding a matching window on any view in the binocular image, continuously changing parallax to obtain at least two corresponding windows corresponding to the matching window in the other view, and determining a crack texture significant coefficient of the matching window according to the image gray scale characteristics of the corresponding region of the matching window and the distance between the center of the matching window and the crack region; carrying out morphological refinement operation and branch point analysis on the crack region, and constructing crack branch point extensibility of a branch point on the edge line of the crack region; identifying corner points at the intersections of the cracks in the matching window, and analyzing by combining the corner points and the extensibility of the branching points of the cracks to determine the complexity coefficients of the branching of the cracks of the matching window; combining the crack texture significant coefficient and the crack branch complex coefficient to determine a matching cost weight coefficient of each matching window; and weighting matching costs corresponding to the matching windows and corresponding windows in the binocular image based on the matching cost weight coefficient, updating the matching costs, screening out optimal parallax and a corresponding parallax map according to the continuously updated matching costs, and determining the crack depth of the bridge tunnel based on the parallax map. According to the method, the potential risk of the bridge can be estimated by measuring the cracking depth through the stereo matching algorithm according to the binocular image of the bridge tunnel shot by the binocular stereo camera, the defect that the accuracy of the traditional detection method is low under the condition of insufficient texture and illumination is overcome, and the reliability of depth detection is improved.
It should be noted that: the sequence of the embodiments of the present invention is only for description, and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments.
Claims (10)
1. The method for measuring the depth of the crack of the bridge tunnel is characterized by comprising the following steps of:
Obtaining binocular images of a bridge tunnel; acquiring a crack region from the binocular image;
sliding a matching window on any view in the binocular image, continuously changing parallax to obtain at least two corresponding windows corresponding to the matching window in the other view, and determining a crack texture significant coefficient of the matching window according to the image gray scale characteristics of the corresponding region of the matching window and the distance between the center of the matching window and the crack region;
Carrying out morphological refinement operation and branch point analysis on the crack region, and constructing crack branch point extensibility of a branch point on the edge line of the crack region; identifying corner points at the intersections of the cracks in the matching window, and analyzing by combining the corner points and the extensibility of the branching points of the cracks to determine the complexity coefficients of the branching of the cracks of the matching window;
Combining the crack texture significant coefficient and the crack branch complex coefficient to determine a matching cost weight coefficient of each matching window; and weighting matching costs corresponding to the matching windows and corresponding windows in the binocular image based on the matching cost weight coefficient, updating the matching costs, screening out optimal parallax and a corresponding parallax map according to the continuously updated matching costs, and determining the crack depth of the bridge tunnel based on the parallax map.
2. The method for measuring the depth of a crack in a bridge tunnel according to claim 1, further comprising, after the step of obtaining the crack region from the binocular image:
and analyzing the gray level of the crack region to determine a deep crack region and a shallow crack region.
3. The method for measuring the depth of a crack in a bridge tunnel according to claim 2, wherein the determining the crack texture saliency coefficient of the matching window according to the image gray scale characteristics of the corresponding area of the matching window and the distance between the center of the matching window and the crack area comprises:
Calculating the inverse variance and contrast of the gray level co-occurrence matrix of the corresponding region of the matching window; and determining the crack texture significant coefficient of the matching window by combining the inverse variance, the contrast, the number of pixels in the deep crack region, the number of pixels in the shallow crack region and the distances between the center of the matching window and the edge lines of the deep crack region and the shallow crack region.
4. The method for measuring the depth of a crack in a bridge tunnel according to claim 3, wherein the calculation formula of the crack texture significant coefficient is as follows:;
Soct is the crack texture significant coefficient of the matching window; CON is the contrast of the gray level co-occurrence matrix of the corresponding region of the matching window; h is the inverse variance of the gray level co-occurrence matrix of the region corresponding to the matching window; nd is the number of pixel points in the deep crack area; ns is the number of pixel points in the shallow crack region; ds is the shortest distance between the center of the matching window and the edge line of the nearest shallow crack region; dd is the shortest distance between the center of the matching window and the edge line of the nearest deep crack region; norm is a normalization function.
5. The method for measuring the depth of a crack in a bridge tunnel according to claim 2, wherein the determining the deep crack region and the shallow crack region comprises:
Clustering crack areas based on gray values to obtain two categories;
calculating the gray average value of the crack area in the two categories;
Taking a crack region in the class with the smallest gray average value in the two classes as a deep crack region; and taking a crack region in the category with the maximum gray average value in the two categories as a shallow crack region.
6. The method for measuring the depth of a crack in a bridge tunnel according to claim 1, wherein the performing morphological refinement operation and branch point analysis on the crack region to construct the crack branch point extensibility of the branch point on the edge line of the crack region comprises:
carrying out morphological refining operation on the crack region, and refining the crack region into crack lines;
Taking a crack line corresponding to a crack region with the largest width as a first-stage crack; acquiring branch points of other cracks on the first-stage crack, marking the branch points as first-stage branch points, and taking crack lines extending from the first-stage branch points as second-stage cracks; acquiring branch points of other cracks on the secondary cracks, marking the branch points as secondary branch points, and taking crack lines extending from the secondary branch points as tertiary cracks; and so on until there are no branching points and cracks;
And for the crack lines corresponding to any crack region, determining the crack branch point extensibility of the branch point by combining the included angle between the crack lines and the crack lines extending at each branch point on the crack lines and the number of cracks at the same branch point, wherein the included angle between the crack lines extending at each branch point on the crack lines and the number of cracks at the same branch point are in positive correlation with the crack branch point extensibility.
7. The method for measuring the depth of a crack in a bridge tunnel according to claim 1, wherein the analyzing the joint corner and the crack branch point extension to determine the crack branch complexity coefficient of the matching window comprises:
clustering the corner points in the matching window to obtain a cluster;
when the corner points are branch points, taking the extension degree of the crack branch points corresponding to the corner points as a molecule, taking the sum value of Euclidean distances between two corner points in a cluster to which the corner points belong as a denominator, obtaining a first ratio, and taking the first ratio as the branch point density degree of the corner points;
and taking the sum of the branch point densities corresponding to all the corner points as a crack branch complexity coefficient of the matching window.
8. The method for measuring the depth of a crack in a bridge tunnel according to claim 1, wherein the determining the matching cost weight coefficient of each matching window by combining the crack texture saliency coefficient and the crack branching complexity coefficient comprises:
Taking the preset parameters as molecules, taking the sum of the crack texture significant coefficient of each matching window and the crack branch complex coefficient of each matching window as denominators to obtain a second ratio, and carrying out inverse proportion normalization on the second ratio to obtain a matching cost weight coefficient of each matching window; wherein the preset parameter is a positive number.
9. The bridge tunnel crack depth measurement method according to claim 1, wherein the calculation formula of the updated matching cost is:;
wherein C (p, q) is the updated matching cost; Matching the cost weight coefficient; is a Census-based cost; At the cost of AD.
10. The method for measuring the depth of a crack in a bridge tunnel according to claim 1, wherein the step of acquiring the crack region from the binocular image comprises the steps of:
crack regions in the binocular image are identified by Niblack thresholding.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410384276.8A CN117974753B (en) | 2024-04-01 | 2024-04-01 | Bridge tunnel crack depth measurement method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410384276.8A CN117974753B (en) | 2024-04-01 | 2024-04-01 | Bridge tunnel crack depth measurement method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN117974753A true CN117974753A (en) | 2024-05-03 |
CN117974753B CN117974753B (en) | 2024-06-14 |
Family
ID=90866124
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202410384276.8A Active CN117974753B (en) | 2024-04-01 | 2024-04-01 | Bridge tunnel crack depth measurement method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117974753B (en) |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106910187A (en) * | 2017-01-13 | 2017-06-30 | 陕西师范大学 | A kind of artificial amplification method of image data set for Bridge Crack detection |
CN110148181A (en) * | 2019-04-25 | 2019-08-20 | 青岛康特网络科技有限公司 | A kind of general binocular solid matching process |
CN110544251A (en) * | 2019-09-08 | 2019-12-06 | 刘凡 | Dam crack detection method based on multi-migration learning model fusion |
CN113506257A (en) * | 2021-07-02 | 2021-10-15 | 同济大学 | Crack extraction method based on self-adaptive window matching |
CN116051992A (en) * | 2023-01-09 | 2023-05-02 | 郑旺 | Crack identification method |
CN117474870A (en) * | 2023-11-03 | 2024-01-30 | 中国公路工程咨询集团有限公司 | Road pavement crack identification decision-making method and system based on big data screening |
-
2024
- 2024-04-01 CN CN202410384276.8A patent/CN117974753B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106910187A (en) * | 2017-01-13 | 2017-06-30 | 陕西师范大学 | A kind of artificial amplification method of image data set for Bridge Crack detection |
CN110148181A (en) * | 2019-04-25 | 2019-08-20 | 青岛康特网络科技有限公司 | A kind of general binocular solid matching process |
CN110544251A (en) * | 2019-09-08 | 2019-12-06 | 刘凡 | Dam crack detection method based on multi-migration learning model fusion |
CN113506257A (en) * | 2021-07-02 | 2021-10-15 | 同济大学 | Crack extraction method based on self-adaptive window matching |
CN116051992A (en) * | 2023-01-09 | 2023-05-02 | 郑旺 | Crack identification method |
CN117474870A (en) * | 2023-11-03 | 2024-01-30 | 中国公路工程咨询集团有限公司 | Road pavement crack identification decision-making method and system based on big data screening |
Also Published As
Publication number | Publication date |
---|---|
CN117974753B (en) | 2024-06-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN115311292B (en) | Strip steel surface defect detection method and system based on image processing | |
CN115829883B (en) | Surface image denoising method for special-shaped metal structural member | |
CN115861291B (en) | Chip circuit board production defect detection method based on machine vision | |
CN114219805B (en) | Intelligent detection method for glass defects | |
CN111754472A (en) | Pulmonary nodule detection method and system | |
CN112288758B (en) | Infrared and visible light image registration method for power equipment | |
CN116385450B (en) | PS sheet wear resistance detection method based on image processing | |
CN116309757B (en) | Binocular stereo matching method based on machine vision | |
CN111242026B (en) | Remote sensing image target detection method based on spatial hierarchy perception module and metric learning | |
CN116137036B (en) | Gene detection data intelligent processing system based on machine learning | |
CN111681275B (en) | Double-feature-fused semi-global stereo matching method | |
CN115457044B (en) | Pavement crack segmentation method based on class activation mapping | |
CN118015002B (en) | Traffic engineering road condition visual detection method and system | |
Haq et al. | An edge-aware based adaptive multi-feature set extraction for stereo matching of binocular images | |
CN116990323A (en) | High-precision printing plate visual detection system | |
CN110348307B (en) | Path edge identification method and system for crane metal structure climbing robot | |
CN117974753B (en) | Bridge tunnel crack depth measurement method | |
CN105809657A (en) | Angular point detection method and device | |
CN100371944C (en) | Greyscale image partition method based on light distribution character of reflection or penetrance | |
CN113284158B (en) | Image edge extraction method and system based on structural constraint clustering | |
CN114399532A (en) | Camera position and posture determining method and device | |
Erukala Mahender et al. | Multi-level thresholding for image segmentation on medical images using multi otsu and sine cosine optimization algorithm | |
CN113744324A (en) | Stereo matching method combining multiple similarity measures | |
CN117455778B (en) | Optical glass prefabricated member detection method and system based on image enhancement | |
Zhang et al. | Insights into local stereo matching: Evaluation of disparity refinement approaches |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |