Pavement defect visual detection method and device applied to highway engineering

Abstract

The invention relates to the technical field of image processing, in particular to a pavement defect visual detection method and device applied to highway engineering, and the method comprises the steps: carrying out the image collection of a preset highway region, and obtaining a pavement detection image; performing gray processing on the road surface detection image to obtain a processed gray image; according to a preset bidirectional edge detection algorithm, carrying out image processing on the grayscale image to obtain an edge detection image; determining at least one pair of target crack areas according to the edge detection image; according to the target crack area, the damage degree of the preset road area is determined, and the damage degree is used for determining the maintenance result of the preset road area. According to the method, the target crack area is obtained by performing image processing and crack identification on the road surface detection image, interference of an area similar to the shape of the target crack area is eliminated, and the precision of road surface crack extraction by a bidirectional edge detection algorithm is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing operations, and in particular to a road surface defect visual detection method and device applied to highway engineering. Background Art

[0002] In highway engineering, the detection of road surface defects is an important part of ensuring road safety and extending service life. At present, the commonly used road surface defect detection technologies mainly include image shooting and radar measurement. These technologies can help engineers to find road surface cracks, subsidence and other defects to a large extent.

[0003] However, existing technologies have certain limitations in their application. Especially in image processing, existing highway crack recognition technologies often rely on bidirectional edge detection algorithms to identify road cracks. When processing actual road scenes, bidirectional edge detection algorithms often encounter repaired crack areas. Since the repaired crack areas are morphologically similar to natural crack areas, this brings great interference to crack recognition. Bidirectional edge detection algorithms are prone to mistakenly identify these repaired crack areas as crack areas during the recognition process, thereby reducing the accuracy and reliability of detection. This misjudgment not only affects road maintenance decisions, but may also lead to waste of resources and inefficiency in maintenance work.

[0004] Therefore, the existing highway pavement defect detection technology needs to be improved urgently. Summary of the invention

[0005] In order to solve the technical problem that the crack area after repair is similar to the crack area to be repaired in the road scene, resulting in the inability to accurately and clearly obtain the crack area to be repaired, the purpose of the present invention is to provide a road surface defect visual detection method and device for highway engineering, and the technical solution adopted is as follows: In a first aspect, an embodiment of the present invention provides a road surface defect visual detection method applied to highway engineering, comprising: Capture images of a preset highway area to obtain a road surface detection image; Performing grayscale processing on the road surface detection image to obtain a processed grayscale image; According to a preset bidirectional edge detection algorithm, the grayscale image is processed to obtain an edge detection image; Identify the edge detection image to obtain multiple first edge areas; perform matching processing on the multiple first edge areas to obtain at least one pair of second edge areas, each pair of second edge areas includes two first edge areas; perform similarity processing on the second edge areas to determine whether each pair of second edge areas is the same area; if they are the same area, perform crack image recognition processing on each pair of second edge areas in the same area to obtain at least one pair of first crack edge areas, the first crack edge areas are edge areas that meet preset crack specifications; perform width change calculation on the first crack edge areas to obtain a set of width change rates of each pair of first crack edge areas in the first crack edge areas, the set of width change rates of each pair of first crack edge areas includes multiple widths and width change rates between the multiple widths; determine the first crack probability of each pair of first crack edge areas based on the set of width change rates of each pair of first crack edge areas; screen the first crack probability of each pair of first crack edge areas to obtain at least one pair of second crack edge areas that are greater than the first preset crack probability; determine multiple target crack areas based on the second crack edge areas; The damage degree of the preset highway area is determined according to the target crack area, and the damage degree is used to determine the maintenance result of the preset highway area.

[0006] In a second aspect, an embodiment of the present invention provides a road surface defect visual detection device applied to highway engineering, the device comprising: The acquisition unit is used to acquire images of a preset highway area to obtain a road surface detection image; A processing unit, used for performing grayscale processing on the road surface detection image to obtain a processed grayscale image; The processing unit is further used to perform image processing on the grayscale image according to a preset bidirectional edge detection algorithm to obtain an edge detection image; A determination unit is used to identify the edge detection image to obtain multiple first edge areas; perform matching processing on the multiple first edge areas to obtain at least one pair of second edge areas, each pair of second edge areas includes two first edge areas; perform similarity processing on the second edge areas to determine whether each pair of second edge areas is the same area; if they are the same area, perform crack image recognition processing on each pair of second edge areas in the same area to obtain at least one pair of first crack edge areas, the first crack edge areas are edge areas that meet preset crack specifications; perform width change calculation on the first crack edge areas to obtain a set of width change rates of each pair of first crack edge areas in the first crack edge areas, the set of width change rates of each pair of first crack edge areas includes multiple widths and width change rates between the multiple widths; determine the first crack probability of each pair of first crack edge areas based on the set of width change rates of each pair of first crack edge areas; screen the first crack probability of each pair of first crack edge areas to obtain at least one pair of second crack edge areas greater than the first preset crack probability; determine multiple target crack areas based on the second crack edge areas; The determination unit is further used to determine the damage degree of the preset highway area according to the target crack area, and the damage degree is used to determine the maintenance result of the preset highway area.

[0007] In a third aspect, a computer device is provided, comprising a memory and a processor, wherein the memory is connected to the processor, and the processor is used to execute one or more computer programs stored in the memory. When the processor executes the one or more computer programs, the computer device implements the visual detection method for pavement defects applied to highway engineering as described in the first aspect.

[0008] In a fourth aspect, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a processor, the processor executes the visual detection method for pavement defects applied to highway engineering as described in the first aspect.

[0009] The present invention has the following beneficial effects: the present invention can quickly cover a large area of ​​highway through image acquisition technology, and compared with manual inspection, the speed and efficiency of detection are greatly improved; the application of grayscale processing and bidirectional edge detection algorithm realizes the automation of the detection process, reduces human intervention, and reduces false detection and missed detection caused by human factors, that is, it can more accurately identify and extract the target crack area in the road surface, and effectively distinguish between real cracks and repair areas; after further determining the target crack area, the degree of damage of the highway can be quantitatively analyzed, thereby providing reliable data support for subsequent maintenance decisions, and through accurate assessment of the degree of damage, a more reasonable maintenance plan can be formulated, resource allocation can be optimized, and unnecessary maintenance costs can be reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are 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.

[0011] Figure 1 A schematic flow chart of a method for visually detecting road surface defects applied to highway engineering provided by one embodiment of the present invention; Figure 2 A schematic diagram of the structure of a road surface defect visual detection device applied to highway engineering provided by one embodiment of the present invention; Figure 3 A schematic diagram of the structure of a computer device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0012] 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 a method and device for visual detection of road surface defects applied to highway engineering according to the present invention, its specific implementation, structure, features and effects, in conjunction 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.

[0013] 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.

[0014] A specific scheme of a pavement defect visual detection device and method for highway engineering provided by the present invention will be described in detail below in conjunction with the accompanying drawings.

[0015] See also Figure 1 , which shows a flow chart of a method for visually detecting road surface defects applied to highway engineering provided by an embodiment of the present invention, the method comprising the following steps: S10: Capture images of a preset highway area to obtain a road surface detection image.

[0016] The preset highway area refers to a road section that is predetermined to be inspected according to a highway maintenance plan or demand. The preset highway area may be selected based on historical data, road condition surveys or regular inspection results.

[0017] Image acquisition refers to the use of a camera or similar acquisition device to capture images of the road surface. The acquisition device can be a professional camera installed on a vehicle, or a photography system carried by a drone. Specifically, when performing image acquisition, the vehicle's driving speed is usually recommended to be within the range of 20-60 km / h. This speed range can improve acquisition efficiency while ensuring image clarity. Adjust the shooting frame rate (such as 30 fps) according to the driving speed to ensure that no image is missed.

[0018] Furthermore, the collected image data will be transmitted in real time or stored in the vehicle's hard disk. Each frame of the image is accompanied by metadata such as timestamp, geographic location information, and vehicle speed.

[0019] Optionally, when capturing images while the vehicle is moving, it is necessary to ensure that the image capture is synchronized with the vehicle speed, and use a vibration reduction device or image stabilization technology to reduce motion blur.

[0020] The road surface detection image refers to the road surface image obtained by the image acquisition device for subsequent analysis. The road surface detection image needs to have sufficient resolution and clarity to accurately identify the tiny defects of the road surface.

[0021] It can be seen that the accuracy and reliability of image acquisition are ensured in this embodiment, which helps to improve the efficiency and accuracy of the road surface defect visual detection method.

[0022] S20, performing grayscale processing on the road surface detection image to obtain a processed grayscale image.

[0023] Grayscale processing refers to the process of converting the color information of each pixel in a color image into a grayscale value. The purpose of grayscale processing is to simplify the image data structure, highlight the important information in the image, and provide a better basis for subsequent edge detection and crack identification.

[0024] Specifically, grayscale images have only one channel, unlike color images which have three channels (red, green, and blue). The image obtained after grayscale processing only contains grayscale information. The pixel values ​​are usually between 0 and 255, where 0 represents black, 255 represents white, and the values ​​in between represent different gray levels.

[0025] Specifically, grayscale processing methods may include but are not limited to the average method, the weighted method (taking into account the sensitivity of the human eye to different colors) and the single channel method (using only a certain channel of the color image). The weighted method generally uses the following formula: grayscale value = 0.299R + 0.587G + 0.114B, where R, G, and B are the pixel values ​​of the red, green, and blue channels, respectively.

[0026] Optionally, during the grayscale processing, it is necessary to maintain the details and contrast of the image so that the subsequent defect detection is not affected, and therefore the grayscale image is enhanced, such as histogram equalization, to improve the visual effect and contrast of the image.

[0027] It can be seen that in this embodiment, through effective grayscale processing, image information can be simplified, image details can be enhanced, and a reliable data basis can be provided for the road surface defect visual detection method.

[0028] S30. Perform image processing on the grayscale image according to a preset bidirectional edge detection algorithm to obtain an edge detection image.

[0029] The preset bidirectional edge detection algorithm is an image processing algorithm that simultaneously detects rising edges and falling edges in an image. A rising edge refers to a transition from dark to light, while a falling edge refers to a transition from light to dark.

[0030] Among them, the edge detection image is that the edge of the object is highlighted in the form of bright lines or specific marks, while other areas are relatively dark or remain unchanged.

[0031] Specifically, the preset bidirectional edge detection algorithm usually uses a gradient operator, such as the Sobel operator or the Canny operator, to calculate the gradient magnitude and direction of each pixel in the image. The Sobel operator highlights the edge by calculating the spatial gradient of the image brightness. The Canny operator is a multi-stage algorithm that includes gradient calculation, non-maximum suppression, and dual threshold edge tracking. Apply the gradient operator to the grayscale image to calculate the gradient of each pixel. Determine the gradient direction and determine whether it is a rising or falling edge. Apply a threshold to determine which gradient values ​​are large enough to be marked as an edge. A threshold that is too low may cause too much noise to be marked as an edge, while a threshold that is too high may cause the real edge to be missed. Depending on the specific conditions of the road surface and the image quality, the parameters of the algorithm, such as the size of the gradient operator, the threshold, etc., may need to be adjusted.

[0032] It can be seen that in this embodiment, by presetting the bidirectional edge detection, the edge information of the road surface defects can be accurately extracted, thereby improving the reliability and efficiency of the road surface defect visual detection method.

[0033] S40: Determine at least one pair of target crack regions according to the edge detection image.

[0034] The target crack area is a crack area that actually exists and has not been repaired in the preset highway area. The crack may appear on both sides of the road surface, forming a pair. The target crack area includes at least two corresponding crack edge areas, which can be a pair of symmetrical crack edges or a continuous crack segment.

[0035] Optionally, identification of target crack regions is achieved by analyzing continuous edge lines in the edge detection image. The algorithm looks for interruptions, bifurcations, or specific shape patterns in edge lines, which are often characteristic of cracks. Morphological operations such as dilation and erosion can be used to connect broken edges and remove noise to more accurately identify crack regions.

[0036] Furthermore, in actual operation, the repaired crack area may show different characteristics in the image, such as uniformity of color, texture or shape, and it is necessary to be able to distinguish these characteristics to avoid misidentifying the repaired area as the target crack area. Therefore, the machine learning model can be trained to identify the difference between the repair material and the natural road surface, thereby improving the accuracy of recognition.

[0037] The specific implementation process of S40 may refer to the specific description of S101-S108, which will not be repeated here.

[0038] It can be seen that in this embodiment, by analyzing the edge detection image, the target crack area can be effectively determined, the probability of misidentifying the repaired crack area as the target crack area is reduced, and the reliability and accuracy of the pavement defect visual detection method are improved.

[0039] S50. Determine the damage degree of the preset highway area according to the target crack area, wherein the damage degree is used to determine the maintenance result of the preset highway area.

[0040] Among them, the degree of damage is a quantitative assessment of the quality of the road surface, reflecting the health and service life of the road surface. It is usually based on the number, length, width, depth and distribution of cracks.

[0041] Among them, the maintenance result refers to the maintenance measures determined according to the degree of damage, including whether maintenance is required, the maintenance method (such as crack filling, milling and resurfacing, etc.), the urgency of maintenance, and the expected maintenance effect. The evaluation result of the degree of damage will be directly used to make maintenance decisions. For example, if the degree of damage is low, only simple crack filling may be required; if the degree of damage is high, more thorough pavement resurfacing may be required.

[0042] Optionally, the length (L) of each crack in all target crack regions of at least one pair of target crack regions is statistically counted, and the total number of cracks (M) is statistically counted; the length (L) of each crack and the total number of cracks (M) are calculated according to the following formula to obtain the degree of damage of the preset highway region; the preset degree of damage is obtained; the degree of damage of the preset highway region is compared with the preset degree of damage to obtain the maintenance result of the preset highway region.

[0043] Specifically, the formula is as follows: In this formula, represents the degree of damage of the currently detected preset highway region. The length of the cracks in the pavement is L, and the number of cracks is M. The larger k is, the greater the degree of pavement damage. norm is a normalization function used to convert the product of the length and number of cracks into a standardized value of the degree of damage.

[0044] Specifically, the preset degree of damage is the benchmark for evaluating the degree of pavement damage. The preset degree of damage can be set to 0.3, which is not uniquely limited here.

[0045] Among them, if k 0.3, it is determined that the current preset highway region is slightly damaged, and the maintenance measures are relatively simple at this time. First, clean the sundries and dust inside the cracks, and then use high-pressure air for cleaning. Then, heat the crack sealant to enhance its fluidity and evenly fill the cracks to ensure that the cracks are completely filled.

[0046] Among them, if k > 0.3, it is determined that the current preset highway region requires more thorough maintenance. Use a sawing machine to cut out regular edges on both sides of the crack, remove the damaged pavement material, and clean the base. Then, fill with new materials and compact to ensure that the pavement restores its structural integrity and function.

[0047] Therefore, different degrees of pavement damage can be evaluated and processed to ensure the safety and durability of the highway.

[0048] It can be seen that through the analysis of the target crack region in this embodiment, the degree of damage of the preset highway region can be accurately evaluated. This evaluation provides a scientific basis for the maintenance decision of the highway, ensuring the safety, durability and economy of the highway.

[0049] The present invention has the following beneficial effects: the present invention can quickly cover a large area of ​​highway through image acquisition technology, and compared with manual inspection, the speed and efficiency of detection are greatly improved; the application of grayscale processing and bidirectional edge detection algorithm realizes the automation of the detection process, reduces human intervention, and reduces false detection and missed detection caused by human factors, that is, it can more accurately identify and extract the target crack area in the road surface, and effectively distinguish between real cracks and repair areas; after further determining the target crack area, the degree of damage of the highway can be quantitatively analyzed, thereby providing reliable data support for subsequent maintenance decisions, and through accurate assessment of the degree of damage, a more reasonable maintenance plan can be formulated, resource allocation can be optimized, and unnecessary maintenance costs can be reduced.

[0050] S101. In one embodiment, the method of determining a plurality of target crack regions based on the edge detection image comprises: identifying the edge detection image to obtain a plurality of first edge regions; performing matching processing on the plurality of first edge regions to obtain at least one pair of second edge regions, each pair of second edge regions including two first edge regions; performing similarity processing on the second edge regions to determine whether each pair of second edge regions are the same region; if they are the same region, performing crack image recognition processing on each pair of second edge regions in the same region to obtain at least one pair of first crack edge regions, the first crack edge region being an edge region that meets preset crack specifications; and determining a plurality of target crack regions based on the first crack edge regions.

[0051] The specific process of identification may be to extract multiple first edge regions from the image using an edge detection algorithm (such as Canny, Sobel, etc.) The first edge regions represent possible edges in the image, including cracks, road surface markings, etc.

[0052] Matching refers to pairing detected edge regions to identify edges that may correspond to the same crack. For example, a crack may have corresponding edges on both sides of the road surface, and these edges need to be identified and paired. Specifically, matching may involve image processing techniques such as template matching, feature point matching, or deep learning-based pairing algorithms to ensure that the edges on both sides of the same crack are accurately matched.

[0053] Each pair of second edge regions is composed of two first edge regions corresponding to each other in spatial position, and they may belong to two sides of the same crack.

[0054] Among them, similarity processing involves comparing the features of paired edge regions, such as shape, size, direction, and spacing, to determine whether they belong to the same crack. If the features match, they are considered to be edges of the same region. Specifically, similarity processing may include calculating shape descriptors of edge regions, using cross-correlation coefficients or other similarity metrics to evaluate the similarity between edge regions.

[0055] Furthermore, for the edge pairs confirmed to belong to the same area, crack image recognition processing is further performed to more accurately identify the characteristics of the cracks.

[0056] Among them, crack image recognition processing may include analysis of the width, length, direction and shape of the crack, and the use of machine learning models to identify the characteristics of the crack.

[0057] Among them, the preset crack specifications refer to the crack feature standards set in the algorithm, such as minimum width, maximum gap, length threshold, etc., which are used to distinguish between real cracks and other types of edges.

[0058] Among them, the specific process of determining multiple target crack areas based on the first crack edge area can be referred to the specific description of S102-S103, and will not be repeated here.

[0059] Among them, the specific process of matching the multiple first edge regions to obtain at least one pair of second edge regions, each pair of second edge regions includes two first edge regions, can be referred to the specific description of S104, which will not be repeated here.

[0060] For the specific process of performing similarity processing on the second edge regions and determining whether each pair of second edge regions are the same region, reference may be made to the specific description of S105 , which will not be repeated here.

[0061] Among them, if it is the same area, crack image recognition processing is performed on each pair of second edge areas in the same area to obtain at least one pair of first crack edge areas, and the first crack edge area is an edge area that meets the preset crack specifications. For the specific process, please refer to S106 for the specific description, which will not be repeated here.

[0062] It can be seen that in this embodiment, through multi-layer screening and image processing, road cracks can be accurately identified from complex image data, and reliable data support can be provided for subsequent damage assessment and maintenance decisions.

[0063] S102. In one embodiment, the method of determining a plurality of target crack regions based on the first crack edge region comprises: calculating the width change of the first crack edge region to obtain a set of width change rates of each pair of first crack edge regions in the first crack edge region, wherein the set of width change rates of each pair of first crack edge regions comprises a plurality of widths and width change rates between the plurality of widths; determining a first crack probability of each pair of first crack edge regions based on the set of width change rates of each pair of first crack edge regions; screening the first crack probability of each pair of first crack edge regions to obtain at least one pair of second crack edge regions having a crack probability greater than a first preset crack probability; and determining a plurality of target crack regions based on the second crack edge regions.

[0064] The width change calculation is specifically to sample pixel rows along the crack edge and calculate the width difference between adjacent sampling points. The width difference can be used to calculate the change rate.

[0065] Specific, available It represents the relative change rate between the i-th width and the i-1-th width on each pair of first crack edge regions. The larger the value, the greater the change between the two widths.

[0066] Further The sum of the change rates between all n widths and the previous one on each pair of first crack edge regions is calculated. The specific formula is expressed as The sum of the rates of change between all n widths on the first crack edge area and the previous one.

[0067] The width change rate set refers to the change rate of the width at different positions for each pair of first crack edge regions. The width change rate reflects the irregularity of the crack width and the true characteristics of the crack. The width change rate set contains not only multiple width measurements of the crack edge, but also the relative change rate between these measurements, that is, how the width changes from one point to the next.

[0068] By analyzing the set of width change rates, the probability that each pair of edge regions is a real crack can be estimated. Usually, the width change of a crack will show a certain pattern, while the edge of a non-crack will not.

[0069] Among them, the first preset crack probability is set. Only when the crack probability of a pair of edge areas exceeds this threshold, it will be considered as a potential crack. The setting of the first preset crack probability is based on empirical data or best practices determined through cross-validation. If the first preset crack probability is too high, it may lead to missed detection, while if it is too low, it may lead to false detection. The first preset crack probability can be set to 0.8, which is not a unique limit here.

[0070] Among them, according to the previous screening, the area with a probability greater than the preset crack probability is the second crack edge area, and the second crack edge area is the edge area most likely to represent the real crack.

[0071] Among them, the specific process of determining the first crack probability of each pair of first crack edge regions according to the width change rate set of each pair of first crack edge regions can be referred to the specific description of S107, which will not be repeated here.

[0072] Among them, the specific process of determining multiple target crack regions based on the second crack edge region can be referred to the specific description of S103, and will not be repeated here.

[0073] It can be seen that in this embodiment, real cracks can be accurately identified from potential crack edge areas and accurate data can be provided for further analysis and maintenance decisions. In S102, real cracks are distinguished from other types of image features through width change rate and probability calculation.

[0074] S103. In one embodiment, the determining of multiple target crack regions based on the second crack edge region includes: performing preset image processing on the second crack edge region to obtain a second crack probability corresponding to each pair of second crack edge regions in the second crack edge region; and screening the second crack probability of each pair of second crack edge regions to obtain at least one pair of target crack regions having a crack probability greater than a second preset crack probability.

[0075] Among them, in the specific process of performing preset image processing on the second crack edge area to obtain the second crack probability corresponding to each pair of second crack edge areas in the second crack edge area, reference may be made to the specific description of S108, which will not be repeated here.

[0076] The second crack probability indicates the possibility that each pair of second crack edge regions is a crack.

[0077] Optionally, the calculation of the second crack probability may involve complex algorithms, such as support vector machine (SVM), random forest, neural network, etc., which can output a probability value based on input feature data.

[0078] The second preset crack probability can be determined through experiments and validation data sets. The second preset crack probability should be high enough to avoid false detection (misidentifying a non-crack area as a crack), but not too high to avoid missed detection (missing the real crack area). The second preset crack probability can be set to 0.85, which is not a unique limit here.

[0079] Therefore, through screening, only those edge areas whose second crack probability is greater than the second preset crack probability will be determined as target crack areas.

[0080] It can be seen that this embodiment can accurately identify real cracks from potential crack edge areas and provide reliable data support for road maintenance and repair, which helps to improve detection accuracy and efficiency, reduce human errors, and speed up the maintenance process.

[0081] S104. In one embodiment, matching processing is performed on the multiple first edge areas to obtain at least one pair of second edge areas, including: selecting any one first edge area from the multiple first edge areas to obtain the first edge area to be processed; calculating the edge length of the first edge area to be processed to obtain the first edge length; calculating the edge length of the first edge area except the first edge area to be processed to obtain multiple second edge lengths; performing difference calculation on the first edge length and all the multiple second edge lengths to obtain multiple differences; selecting the minimum difference among the multiple differences as the target difference, and obtaining the target second edge length corresponding to the target difference; determining the first edge area corresponding to the target second edge length as the matching edge area corresponding to the first edge area to be processed; obtaining the second edge area based on the first edge area to be processed and the matching edge area corresponding to the first edge area to be processed; traversing the multiple first edge areas to obtain at least one pair of second edge areas.

[0082] Among the plurality of first edge regions, one edge region is randomly selected as the region to be processed, and this region will be used as a reference for finding a matching edge region.

[0083] The calculation of edge length usually involves image processing techniques, such as contour tracking or edge detection algorithms, to determine the number of pixels on the edge or to calculate the actual distance through the coordinates of the pixels. Therefore, the first edge length may be the number of pixels along the edge or the actual distance.

[0084] The difference calculation is to find the edge area that is closest to the reference edge area length, which helps to identify the edges that may be on both sides of the same crack. Among all the calculated differences, the smallest difference is found, that is, the target difference, which represents another edge area that is closest to the length of the edge area to be processed. Therefore, the second edge length corresponding to the target difference is the target second edge length.

[0085] The first edge region corresponding to the minimum difference is considered as the matching edge region of the edge region to be processed. This matching region is most similar to the region to be processed in length.

[0086] Among them, the traversal process ensures that each edge region is processed and all possible matching pairs are found as much as possible, thereby improving the comprehensiveness and accuracy of detection. Therefore, by traversing all the first edge regions and repeating the above steps, at least one pair of second edge regions is obtained.

[0087] It can be seen that in this embodiment, the edge regions of cracks can be identified and matched from the image, providing a basis for subsequent crack analysis and pavement damage assessment.

[0088] S105. In one embodiment, similarity processing is performed on the second edge regions to determine whether each pair of second edge regions is the same region, including: selecting any pair of second edge regions to obtain the second edge regions to be processed; obtaining the first target direction and the first pixel data of the second edge regions to be processed; performing data processing according to the first target direction and the first pixel data of the second edge regions to be processed to obtain a first chain code numerical sequence and a second chain code numerical sequence, where the first chain code numerical sequence corresponds to the first first edge region in the second edge regions to be processed, and the second chain code numerical sequence corresponds to the second first edge region in the second edge regions to be processed; performing a similarity operation according to the first chain code numerical sequence and the second chain code numerical sequence to obtain a first matching value between the first chain code numerical sequence and the second chain code numerical sequence; calculating multiple second matching values between the first chain code numerical sequence and multiple third chain code numerical sequences other than the second chain code numerical sequence in turn, where the multiple third chain code numerical sequences are obtained from other second edge regions outside the second edge regions to be processed; when the first matching value is greater than all the multiple second matching values, determining that the first first edge region and the second first edge region in the second edge regions to be processed are the same region.

[0089] Among them, the first target direction is to specify a fixed direction (for example, from left to right, from top to bottom) to facilitate the subsequent generation of the chain code numerical sequence. This specification can be set artificially or, and there is no unique limitation here.

[0090] Among them, the second edge regions to be processed may be the two edges of the same crack or potential matching pairs that need to be further verified.

[0091] Among them, the chain code is a coding method used to represent the direction of each pixel point on the edge. The chain code numerical sequence is a boundary description method that can concisely represent the shape and direction of the edge.

[0092] Specifically, the second edge region to be processed is specified with a fixed direction (from left to right), and along the target direction of the second edge region to be processed, starting from the first pixel point, the direction of these pixels is represented by chain code values. Thus, two chain code value sequences of the second edge region to be processed are obtained.

[0093] Specifically, the first chain code value sequence corresponds to the first first edge region in the second edge region to be processed, which can be The second chain code value sequence corresponds to the second first edge region and can be expressed as express.

[0094] The similarity operation is used to determine whether two edge regions are similar enough to be considered as two parts of the same crack. The matching value is a quantitative indicator to measure the similarity of two chain code sequences.

[0095] Specifically, in the process of performing a similarity operation on the first chain code value sequence and the second chain code value sequence to obtain a first matching value between the first chain code value sequence and the second chain code value sequence, the operation may be performed by the following formula: In this formula, represents the possibility that the two first edge regions in the second edge region to be processed are a pair, that is, the first matching value between the first chain code value sequence and the second chain code value sequence, and h represents the number of pixels in the two chain code value sequences; represents the chain code value of the jth pixel point in the chain code value sequence of the first first edge area in the second edge area to be processed, Represents the chain code value of the j-th pixel point in the chain code value sequence of the second first edge area in the second edge area to be processed. It represents the difference in chain code values ​​of the j-th pixel point in the second edge area to be processed. If the difference is smaller, it means that the direction is more similar (the invention uses 8-chain code to represent it, and stipulates that |1-7|=1). It represents the sum of the differences in chain code values ​​between all pixels in the second edge region to be processed and the corresponding pixels in the other edge. The smaller the value, the greater the possibility that the pair of edges are the same crack edge or the repaired crack edge. The larger the value.

[0096] Among them, the matching values ​​between the first chain code value sequence and the chain code value sequences (third chain code value sequences) of other edge regions except the second chain code value sequence are calculated in sequence. These third chain code value sequences come from other second edge regions except the second edge region to be processed. If the first matching value is greater than all the second matching values, it can be considered that the first first edge region and the second first edge region in the second edge region to be processed belong to the same crack region. By comparing the matching values, it can be determined which pair of edge regions is most likely to be caused by the same crack. This method helps to reduce false matches and improve the accuracy of detection.

[0097] It can be seen that in this embodiment, the crack edge can be accurately identified from the image, and the edge areas belonging to the same crack can be matched, providing reliable data support for the damage degree assessment and maintenance of the road surface.

[0098] S106. In one embodiment, if the area is the same area, crack image recognition processing is performed on each pair of second edge areas in the same area to obtain at least one pair of first crack edge areas, including: obtaining the second target direction and second pixel data in each pair of second edge areas in the same area; performing data processing according to the second target direction and second pixel data in each pair of second edge areas in the same area to obtain a third chain code value sequence and a fourth chain code value sequence, wherein the third chain code value sequence and the fourth chain code value sequence respectively represent the first first edge area and the second edge area of ​​each pair of second edge areas in the same area; edge tortuosity calculation is performed on the third chain code value sequence and the fourth chain code value sequence respectively to obtain a first tortuosity corresponding to the third chain code value sequence and a second tortuosity corresponding to the fourth chain code value sequence; when the first tortuosity and the second tortuosity are both greater than the preset tortuosity, each pair of second edge areas in the same area is determined to be an edge area that meets the preset crack specification, and at least one pair of first crack edge areas is obtained, and the first crack edge area is an edge area that meets the preset crack specification.

[0099] The second target direction is to specify a fixed direction (for example, from left to right, from top to bottom) to facilitate the subsequent generation of chain code value sequences. This specification can be set manually or not, and is not limited here.

[0100] Specifically, for each pair of second edge regions in the same region, a fixed direction (from left to right) is specified, and along the target direction of each pair of second edge regions in the same region, starting from the first pixel point, the direction of these pixels is represented by chain code values. Thus, two chain code value sequences for each pair of second edge regions in the same region are obtained. The third chain code value sequence and the fourth chain code value sequence represent the first first edge region and the second edge region in each pair of second edge regions in the same region, respectively.

[0101] Wherein, in the process of respectively calculating the edge tortuosity of the third chain code value sequence and the fourth chain code value sequence to obtain the first tortuosity corresponding to the third chain code value sequence and the second tortuosity corresponding to the fourth chain code value sequence, the tortuosity calculation can be performed by the following formula: In this formula, Indicates the tortuosity of an edge in each pair of second edge regions in the same region, represents a pixel point on this zigzag edge. express The previous pixel adjacent to the pixel; Represents the chain code value of a pixel point on this zigzag edge. Indicates the chain code value of the next pixel in this zigzag edge. Used to determine whether the direction of these two pixels changes. It is an indicator function, which takes 1 when the two pixels are in different directions, and 0 otherwise. Indicates the number of direction changes between all previous and next pixels on this edge. is the total number of pixels on a certain edge in each pair of second edge regions in the same region; It means to normalize this number, then The larger the value, the greater the tortuosity of the edge, which means the edge where the edge is located is more likely to be the crack edge area. Indicates that the number of changes is normalized for comparison under different edge lengths. The preset tortuosity can be set to 0.6, which is not a unique limit here. The preset tortuosity is used to extract all edges with tortuosity greater than the preset tortuosity, which are considered to be edge areas that meet the preset crack specifications.

[0102] Among them, when the first tortuosity and the second tortuosity are both greater than the preset tortuosity threshold, it can be considered that the pair of edge regions meet the preset crack specifications, and thus they are determined to be the first crack edge regions.

[0103] It can be seen that in this embodiment, the crack edges can be accurately identified from the image and distinguished from the edges after filling, which helps to improve the accuracy and efficiency of crack detection.

[0104] S107. In one embodiment, the determining the first crack probability of each pair of first crack edge regions according to the set of width change rates of each pair of first crack edge regions includes: calculating the width change rate between the multiple widths in the set of width change rates of each pair of first crack edge regions to obtain the mean width change rate of each pair of first crack edge regions; and normalizing the mean width change rate of each pair of first crack edge regions to obtain the first crack probability of each pair of first crack edge regions.

[0105] Among them, the mean width change rate reflects the overall trend of the width change of the crack edge. The larger the mean width change rate, the more uneven the change, which may mean a higher severity of the crack. The mean width change rate refers to, for each pair of first crack edge regions, summing all the width change rates and then dividing by the number of change rates (n - 1, where n is the number of widths). This mean represents the average degree of the width change of the crack edge.

[0106] Among them, the normalization process is to convert the mean width change rate into a dimensionless probability value between 0 and 1. This can be achieved by dividing the mean by the possible maximum change rate, or using other normalization techniques such as min - max normalization or Z - score normalization.

[0107] Therefore, the normalized mean width change rate is the first crack probability, which represents the probability that the edge region is a crack. The higher this probability value, the greater the likelihood that the region is a crack.

[0108] Among them, as mentioned in S102, it can be used to represent the relative change rate between the i - th width and the (i - 1) - th width on each pair of first crack edge regions. The larger this value, the greater the change between the two widths. represents the sum of the change rates between all n widths and the previous one on each pair of first crack edge regions.

[0109] Therefore, represents the mean width change rate on each pair of first crack edge regions. The larger this value, the more uneven the width distribution of each pair of first crack edge regions, that is, the greater the likelihood that each pair of first crack edge regions is a crack region.

[0110] Among them, the normalization process can be represented by the following formula: In this formula, represents the probability that each pair of first crack edge regions is the first crack region, It represents the relative change rate between the i-th width and the i-1-th width on each pair of first crack edge regions. The larger the value, the greater the change between the two widths. represents the sum of the change rates between all n widths of each pair of first crack edge regions and the previous one, It represents the average value of the width change rate of each pair of first crack edge regions. The larger the value is, the more uneven the width distribution of each pair of first crack edge regions is, that is, the greater the possibility that each pair of first crack edge regions is a crack region.

[0111] It can be seen that in this embodiment, meaningful features are extracted from image data and converted into actionable indicators, thereby improving the accuracy and efficiency of road surface defect detection.

[0112] S108. In one embodiment, performing preset image processing on the second crack edge region to obtain the probability of the second crack corresponding to each pair of second crack edge regions in the second crack edge region includes: obtaining the edge extension direction of each pair of second crack edge regions in the second crack edge region and the edge width of each pair of second crack edge regions; generating a plurality of target sliding windows corresponding to each pair of second crack edge regions according to the edge extension direction of each pair of second crack edge regions and taking the edge width of each pair of second crack edge regions as the side length, wherein the window side length of the target sliding window changes with the edge extension direction of each pair of second crack edge regions; obtaining the pixel points of each target sliding window in the plurality of target sliding windows corresponding to each pair of second crack edge regions; calculating the pixel points of each target sliding window to obtain the pixel point ratio of each target sliding window; performing probability calculation according to the pixel point ratio of each target sliding window corresponding to each pair of second crack edge regions to obtain the probability of the second crack corresponding to each pair of second crack edge regions.

[0113] The edge extension direction is usually determined by analyzing the direction of the edge points, while the edge width may be obtained by measuring the minimum distance between the two sides of the edge.

[0114] Among them, the generation of the target sliding window is dynamic. It needs to follow the shape and width changes of the edge, which involves the geometric transformation of the image and the region growing algorithm.

[0115] Among them, multiple target sliding windows can be represented by m windows.

[0116] Specifically, a rectangular sliding window is generated along the extension direction of the edge with the edge width as the side length. The size and shape of the sliding window will change along the extension direction of the edge to adapt to the twists and changes of the edge. In this way, multiple (m) sliding windows are obtained.

[0117] Among them, in the specific process of performing probability calculation according to the pixel point proportion of each target sliding window corresponding to each pair of second crack edge regions to obtain the probability of the second crack corresponding to each pair of second crack edge regions, the calculation can be performed by the following formula: In this formula, represents the probability of each pair of second crack edge regions corresponding to the second crack, is the percentage of pixels in a single window among the m windows, It is the average percentage of pixels in all windows among these m windows. The larger the value, the greater the possibility that each pair of second crack edge regions is a crack region.

[0118] The calculation of pixel percentage is usually completed by counting the number of pixels belonging to crack features in the window and then dividing it by the total number of pixels in the window. The pixel percentage reflects the density of crack features in the window.

[0119] It can be seen that the combination of image processing and statistical analysis is utilized in this embodiment, which can more accurately evaluate the authenticity of the suspected crack area and improve the accuracy and automation of crack detection.

[0120] It should be noted that the sequence of the above 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 specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0121] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0122] As another aspect of an embodiment of the present invention, an embodiment of the present invention provides a visual detection device for road surface defects applied to highway engineering. The visual detection device for road surface defects applied to highway engineering can be a software module, which includes several instructions stored in a memory, and a processor can access the memory and call the instructions for execution to complete the visual detection method for road surface defects applied to highway engineering described in the above-mentioned various embodiments.

[0123] See also Figure 2 , Figure 2 FIG. 1 is a schematic diagram of a structure of a road surface defect visual detection device for highway engineering provided by an embodiment of the present invention. Figure 2 As shown, the pavement defect visual detection device 200 applied to highway engineering includes: The acquisition unit 201 is used to acquire images of a preset highway area to obtain a road surface detection image; The processing unit 202 is used to perform grayscale processing on the road surface detection image to obtain a processed grayscale image; The processing unit 202 is further configured to perform image processing on the grayscale image according to a preset bidirectional edge detection algorithm to obtain an edge detection image; A determination unit 203, configured to determine at least one pair of target crack regions according to the edge detection image; The determination unit 203 is further used to determine the damage degree of the preset highway area according to the target crack area, and the damage degree is used to determine the maintenance result of the preset highway area.

[0124] The present invention has the following beneficial effects: the present invention can quickly cover a large area of ​​highway through image acquisition technology, and compared with manual inspection, the speed and efficiency of detection are greatly improved; the application of grayscale processing and bidirectional edge detection algorithm realizes the automation of the detection process, reduces human intervention, and reduces false detection and missed detection caused by human factors, that is, it can more accurately identify and extract the target crack area in the road surface, and effectively distinguish between real cracks and repair areas; after further determining the target crack area, the degree of damage of the highway can be quantitatively analyzed, thereby providing reliable data support for subsequent maintenance decisions, and through accurate assessment of the degree of damage, a more reasonable maintenance plan can be formulated, resource allocation can be optimized, and unnecessary maintenance costs can be reduced.

[0125] In one embodiment, in determining multiple target crack areas based on the edge detection image, the determination unit 203 is further used to: identify the edge detection image to obtain multiple first edge areas; perform matching processing on the multiple first edge areas to obtain at least one pair of second edge areas, each pair of second edge areas includes two first edge areas; perform similarity processing on the second edge areas to determine whether each pair of second edge areas are the same area; if they are the same area, perform crack image recognition processing on each pair of second edge areas in the same area to obtain at least one pair of first crack edge areas, the first crack edge areas are edge areas that meet preset crack specifications; and determine multiple target crack areas based on the first crack edge areas.

[0126] In one embodiment, in determining a plurality of target crack regions based on the first crack edge region, the determination unit 203 is further used to: calculate the width change of the first crack edge region to obtain a set of width change rates of each pair of first crack edge regions in the first crack edge region, wherein the set of width change rates of each pair of first crack edge regions includes a plurality of widths and width change rates between the plurality of widths; determine the first crack probability of each pair of first crack edge regions based on the set of width change rates of each pair of first crack edge regions; screen the first crack probability of each pair of first crack edge regions to obtain at least one pair of second crack edge regions having a crack probability greater than a first preset crack probability; and determine a plurality of target crack regions based on the second crack edge regions.

[0127] In one embodiment, in determining a plurality of target crack regions based on the second crack edge region, the determination unit 203 is further used to: perform preset image processing on the second crack edge region to obtain a second crack probability corresponding to each pair of second crack edge regions in the second crack edge region; and screen the second crack probability of each pair of second crack edge regions to obtain at least one pair of target crack regions having a crack probability greater than a second preset crack probability.

[0128] In one embodiment, in the matching processing of the multiple first edge areas to obtain at least one pair of second edge areas, the determination unit 203 is also used to: select any one first edge area among the multiple first edge areas to obtain the first edge area to be processed; calculate the edge length of the first edge area to be processed to obtain the first edge length; calculate the edge length of the first edge area except the first edge area to be processed to obtain multiple second edge lengths; perform difference calculation between the first edge length and all the multiple second edge lengths to obtain multiple differences; select the minimum difference among the multiple differences as the target difference, and obtain the target second edge length corresponding to the target difference; determine that the first edge area corresponding to the target second edge length is the matching edge area corresponding to the first edge area to be processed; obtain the second edge area based on the first edge area to be processed and the matching edge area corresponding to the first edge area to be processed; traverse the multiple first edge areas to obtain at least one pair of second edge areas.

[0129] In one embodiment, in the similarity processing of the second edge regions, to determine whether each pair of second edge regions are in the same region, the determining unit 203 is further used to: select any pair of second edge regions to obtain a second edge region to be processed; obtain a first target direction and first pixel data of the second edge region to be processed; perform data processing according to the first target direction and the first pixel data of the second edge region to be processed to obtain a first chain code value sequence and a second chain code value sequence, wherein the first chain code value sequence corresponds to the first first edge region in the second edge region to be processed, and the second chain code value sequence corresponds to the second edge region to be processed. a second first edge region in the domain; performing a similarity operation according to the first chain code value sequence and the second chain code value sequence to obtain a first matching value between the first chain code value sequence and the second chain code value sequence; sequentially calculating multiple second matching values ​​between the first chain code value sequence and multiple third chain code value sequences except the second chain code value sequence, wherein the multiple third chain code value sequences are obtained from other second edge regions except the second edge region to be processed; when the first matching values ​​are all greater than the multiple second matching values, determining that the first first edge region and the second first edge region in the second edge region to be processed are the same region.

[0130] In one embodiment, if the two edges are in the same area, crack image recognition processing is performed on each pair of second edge areas in the same area to obtain at least one pair of first crack edge areas, and the determination unit 203 is further used to: obtain the second target direction and second pixel data of each pair of second edge areas in the same area; perform data processing according to the second target direction and second pixel data of each pair of second edge areas in the same area to obtain a third chain code value sequence and a fourth chain code value sequence, wherein the third chain code value sequence and the fourth chain code value sequence respectively represent the first first edge area and the second edge area of ​​each pair of second edge areas in the same area; perform edge tortuosity calculation on the third chain code value sequence and the fourth chain code value sequence respectively to obtain a first tortuosity corresponding to the third chain code value sequence and a second tortuosity corresponding to the fourth chain code value sequence; when the first tortuosity and the second tortuosity are both greater than a preset tortuosity, determine that each pair of second edge areas in the same area are edge areas that meet the preset crack specifications, and obtain at least one pair of first crack edge areas, wherein the first crack edge area is an edge area that meets the preset crack specifications.

[0131] In one embodiment, in determining the first crack probability of each pair of first crack edge regions based on the width change rate set of each pair of first crack edge regions, the determination unit 203 is further used to: calculate the width change rate between the multiple widths in the width change rate set of each pair of first crack edge regions to obtain the average width change rate of each pair of first crack edge regions; and normalize the average width change rate of each pair of first crack edge regions to obtain the first crack probability of each pair of first crack edge regions.

[0132] In one embodiment, in performing preset image processing on the second crack edge region to obtain the probability that each pair of second crack edge regions in the second crack edge region corresponds to the second crack, the determining unit 203 is further used to: obtain the edge extension direction of each pair of second crack edge regions in the second crack edge region and the edge width of each pair of second crack edge regions; generate multiple target sliding windows corresponding to each pair of second crack edge regions according to the edge extension direction of each pair of second crack edge regions and taking the edge width of each pair of second crack edge regions as the side length, wherein the window side length of the target sliding window changes with the edge extension direction of each pair of second crack edge regions; obtain the pixel points of each target sliding window in the multiple target sliding windows corresponding to each pair of second crack edge regions; calculate the pixel points of each target sliding window to obtain the pixel point ratio of each target sliding window; perform probability calculation according to the pixel point ratio of each target sliding window corresponding to each pair of second crack edge regions to obtain the probability that each pair of second crack edge regions corresponds to the second crack.

[0133] It should be noted that the above-mentioned pavement defect visual detection device applied to highway engineering can execute the pavement defect visual detection method applied to highway engineering provided by the embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method. For technical details not fully described in the embodiment of the pavement defect visual detection device applied to highway engineering, please refer to the pavement defect visual detection method applied to highway engineering provided by the embodiment of the present invention.

[0134] See also Figure 3 , Figure 3 Schematic diagram of the structure of a computer device provided by an embodiment of the present invention. Figure 3 As shown, the computer device 300 includes a processor 301 and a memory 302. The processor 301 is in communication connection with the memory 302.

[0135] The processor 301 is configured to support the computer device to execute the corresponding functions in the pavement defect visual detection method applied to highway engineering in the above method embodiment. The processor 301 can be a central processing unit (CPU), a network processor (NP), a hardware chip or any combination thereof. The above hardware chip can be an application specific integrated circuit (ASIC), a programmable logic device (PLD) or a combination thereof. The above PLD can be a complex programmable logic device (CPLD), a field-programmable gate array (FPGA), a generic array logic (GAL) or any combination thereof.

[0136] Specifically, the processor 301 may include a sending card, a receiving card and a driving chip.

[0137] The memory 302 is used to store program codes, etc. The memory 302 may include a volatile memory (VM), such as a random access memory (RAM); the memory 302 may also include a non-volatile memory (NVM), such as a read-only memory (ROM), a flash memory, a hard disk drive (HDD) or a solid-state drive (SSD); the memory 302 may also include a combination of the above-mentioned types of memory.

[0138] An embodiment of the present invention also provides a computer-readable storage medium, which stores a computer program. The computer program includes program instructions, and when the program instructions are executed by a computer, the computer executes the visual detection method for road surface defects applied to highway engineering as described in the above-mentioned embodiment.

[0139] A person skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

Claims

1. A road surface defect visual detection method applied to highway engineering, characterized in that: include: Capture images of a preset highway area to obtain a road surface detection image; Performing grayscale processing on the road surface detection image to obtain a processed grayscale image; According to a preset bidirectional edge detection algorithm, the grayscale image is processed to obtain an edge detection image; Identify the edge detection image to obtain multiple first edge areas; perform matching processing on the multiple first edge areas to obtain at least one pair of second edge areas, each pair of second edge areas includes two first edge areas; perform similarity processing on the second edge areas to determine whether each pair of second edge areas is the same area; if they are the same area, perform crack image recognition processing on each pair of second edge areas in the same area to obtain at least one pair of first crack edge areas, the first crack edge areas are edge areas that meet preset crack specifications; perform width change calculation on the first crack edge areas to obtain a set of width change rates of each pair of first crack edge areas in the first crack edge areas, the set of width change rates of each pair of first crack edge areas includes multiple widths and width change rates between the multiple widths; determine the first crack probability of each pair of first crack edge areas based on the set of width change rates of each pair of first crack edge areas; screen the first crack probability of each pair of first crack edge areas to obtain at least one pair of second crack edge areas that are greater than the first preset crack probability; determine multiple target crack areas based on the second crack edge areas; The damage degree of the preset highway area is determined according to the target crack area, and the damage degree is used to determine the maintenance result of the preset highway area.

2. The method for visually detecting road surface defects in highway engineering according to claim 1, characterized in that: Determining a plurality of target crack regions according to the second crack edge region includes: Performing preset image processing on the second crack edge region to obtain a second crack probability corresponding to each pair of second crack edge regions in the second crack edge region; The second crack probability of each pair of second crack edge regions is screened to obtain at least one pair of target crack regions having a crack probability greater than a second preset crack probability.

3. The method for visually detecting road surface defects applied to highway engineering according to claim 1, characterized in that: The matching process is performed on the plurality of first edge regions to obtain at least one pair of second edge regions, including: Select any one first edge region from the plurality of first edge regions to obtain a first edge region to be processed; Calculating the edge length of the first edge area to be processed to obtain a first edge length; Calculating edge lengths of first edge regions other than the first edge region to be processed to obtain a plurality of second edge lengths; Calculate the difference between the first edge length and all the second edge lengths to obtain a plurality of differences; Selecting a minimum difference value from the multiple differences as a target difference value, and obtaining a target second edge length corresponding to the target difference value; Determine that the first edge region corresponding to the target second edge length is the matching edge region corresponding to the first edge region to be processed; Obtaining a second edge region according to the first edge region to be processed and a matching edge region corresponding to the first edge region to be processed; The plurality of first edge regions are traversed to obtain at least one pair of second edge regions.

4. The method for visually detecting road surface defects in highway engineering according to claim 1, characterized in that: The performing similarity processing on the second edge regions to determine whether each pair of second edge regions are the same region includes: Selecting any pair of second edge regions to obtain a second edge region to be processed; Acquire a first target direction and first pixel data of the second edge area to be processed; Performing data processing according to the first target direction of the second edge region to be processed and the first pixel data to obtain a first chain code value sequence and a second chain code value sequence, wherein the first chain code value sequence corresponds to the first first edge region in the second edge region to be processed, and the second chain code value sequence corresponds to the second first edge region in the second edge region to be processed; Performing a similarity operation on the first chain code value sequence and the second chain code value sequence to obtain a first matching value between the first chain code value sequence and the second chain code value sequence; sequentially calculating a plurality of second matching values ​​between the first chain code value sequence and a plurality of third chain code value sequences other than the second chain code value sequence, wherein the plurality of third chain code value sequences are obtained from other second edge regions other than the second edge region to be processed; When all of the first matching values ​​are greater than the plurality of second matching values, it is determined that the first first edge region and the second first edge region in the second edge region to be processed are the same region.

5. The method for visually detecting road surface defects in highway engineering according to claim 1, characterized in that: If the area is the same area, crack image recognition processing is performed on each pair of second edge areas in the same area to obtain at least one pair of first crack edge areas, including: Acquire the second target direction and second pixel data in each pair of second edge areas in the same area; Performing data processing according to the second target direction and the second pixel data in each pair of second edge regions in the same region, obtaining a third chain code value sequence and a fourth chain code value sequence, wherein the third chain code value sequence and the fourth chain code value sequence respectively represent a first edge region and a second edge region in each pair of second edge regions in the same region; Calculating edge tortuosity for the third chain code value sequence and the fourth chain code value sequence respectively, to obtain a first tortuosity corresponding to the third chain code value sequence and a second tortuosity corresponding to the fourth chain code value sequence; When the first tortuosity and the second tortuosity are both greater than the preset tortuosity, each pair of second edge regions in the same area is determined to be an edge region that meets the preset crack specifications, and at least one pair of first crack edge regions is obtained, and the first crack edge region is an edge region that meets the preset crack specifications.

6. The method for visually detecting road surface defects in highway engineering according to claim 1, characterized in that: Determining the first crack probability of each pair of first crack edge regions according to the width change rate set of each pair of first crack edge regions includes: Calculating the width change rates between the multiple widths in the width change rate set of each pair of first crack edge regions to obtain an average value of the width change rates of each pair of first crack edge regions; The average value of the width change rate of each pair of first crack edge regions is normalized to obtain the first crack probability of each pair of first crack edge regions.

7. The method for visually detecting road surface defects applied to highway engineering according to claim 2, characterized in that: The performing preset image processing on the second crack edge region to obtain the second crack probability corresponding to each pair of second crack edge regions in the second crack edge region includes: Obtaining the edge extension direction of each pair of second crack edge regions in the second crack edge region and the edge width of each pair of second crack edge regions; According to the edge extension direction of each pair of second crack edge regions, a plurality of target sliding windows corresponding to each pair of second crack edge regions are generated with the edge width of each pair of second crack edge regions as the side length, wherein the window side length of the target sliding window changes with the edge extension direction of each pair of second crack edge regions; Acquire pixel points of each target sliding window in a plurality of target sliding windows corresponding to each pair of second crack edge regions; Calculating the pixel points of each target sliding window to obtain the pixel point ratio of each target sliding window; A probability calculation is performed based on the pixel percentage of each target sliding window corresponding to each pair of second crack edge regions to obtain the second crack probability corresponding to each pair of second crack edge regions.

8. A road surface defect visual detection device used in highway engineering, characterized in that: The device comprises: The acquisition unit is used to acquire images of a preset highway area to obtain a road surface detection image; A processing unit, used for performing grayscale processing on the road surface detection image to obtain a processed grayscale image; The processing unit is further used to perform image processing on the grayscale image according to a preset bidirectional edge detection algorithm to obtain an edge detection image; A determination unit is used to identify the edge detection image to obtain multiple first edge areas; perform matching processing on the multiple first edge areas to obtain at least one pair of second edge areas, each pair of second edge areas includes two first edge areas; perform similarity processing on the second edge areas to determine whether each pair of second edge areas is the same area; if they are the same area, perform crack image recognition processing on each pair of second edge areas in the same area to obtain at least one pair of first crack edge areas, the first crack edge areas are edge areas that meet preset crack specifications; perform width change calculation on the first crack edge areas to obtain a set of width change rates of each pair of first crack edge areas in the first crack edge areas, the set of width change rates of each pair of first crack edge areas includes multiple widths and width change rates between the multiple widths; determine the first crack probability of each pair of first crack edge areas based on the set of width change rates of each pair of first crack edge areas; screen the first crack probability of each pair of first crack edge areas to obtain at least one pair of second crack edge areas greater than the first preset crack probability; determine multiple target crack areas based on the second crack edge areas; The determination unit is further used to determine the damage degree of the preset highway area according to the target crack area, and the damage degree is used to determine the maintenance result of the preset highway area.

Images