Asphalt paving flatness analysis method and system

Abstract

The invention relates to the field of image processing, in particular to a flatness analysis method and system for asphalt paving, and the method comprises the steps: obtaining a preprocessed surface grey-scale map of an asphalt paving pavement, and carrying out the edge detection of the surface grey-scale map to obtain an edge; the edge evaluation of any edge is calculated, the edge with the edge evaluation larger than a preset evaluation threshold value is reserved as a to-be-detected edge, the judgment result of any to-be-detected edge is determined, the judgment result is a real defect area and a non-defect area, and flatness analysis is completed according to the judgment result. According to the technical scheme, the accuracy and efficiency of the flatness analysis result can be improved.

Description

Technical Field

[0001] This invention relates to the field of image processing. In particular, it relates to a method and system for analyzing the smoothness of asphalt paving. Background Technology

[0002] Asphalt paving technology, as a key component of road construction, is widely used globally in the construction and maintenance of various transportation facilities, including highways, urban roads, and airport runways. Asphalt mixtures are laid onto the road surface using a paver, and subsequent compaction and cooling form a robust road structure.

[0003] Smoothness is one of the core indicators for evaluating the quality of asphalt pavement. It is typically assessed by measuring the unevenness of the pavement surface after paving. In current road construction, smoothness monitoring is generally done manually or with simple mechanical measuring tools. However, with the expansion of construction scale and the increasing complexity of the construction environment, traditional manual measurement methods have gradually revealed their drawbacks, such as large measurement errors, low work efficiency, and the inability to obtain smoothness data in real time during construction. This makes it impossible to make timely adjustments during paving, resulting in some sections of the road failing to meet the expected smoothness standards and affecting the overall project quality. Summary of the Invention

[0004] To address the aforementioned technical problems, the present invention provides solutions in the following aspects.

[0005] In the first aspect, a method for analyzing the smoothness of asphalt paving includes: acquiring a grayscale image of the pre-processed asphalt paved surface; performing edge detection on the grayscale image to obtain edges; calculating the edge evaluation of any edge; retaining edges with edge evaluations greater than a preset evaluation threshold as edges to be tested; determining the judgment result of any edge to be tested, wherein the judgment result is a real defect area or a non-defect area; and completing the smoothness analysis based on the judgment result. The calculation of the edge evaluation of any edge includes: taking pixels on the edge as edge points; converting the gradient direction of any edge point into a two-dimensional vector; calculating the mean of the two-dimensional vectors of all edge points in the same dimension; constructing a mean vector; restoring the mean vector to an average angle using an arctangent function; taking the absolute difference between the gradient direction of any edge point and the average angle as the first difference; taking the difference between the full angle and the first difference as the second difference; selecting the minimum value of the first difference and the second difference of any edge point as the evaluation value; calculating the mean of the evaluation values ​​of all edge points; calculating the first ratio of 1 to the right angle; calculating the product of the first ratio and the mean of the evaluation values; and taking the difference between 1 and the product as the edge evaluation.

[0006] Preferably, the determination result of any edge to be measured includes: the edge to be measured includes closed edge to be measured and non-closed edge to be measured; the determination result of any edge to be measured includes the determination result of closed edge to be measured and the determination result of non-closed edge to be measured.

[0007] Preferably, the determination result of the closed test edge includes: calculating the gradient magnitude change of any closed test edge in any preset closure direction, traversing to obtain the gradient magnitude change of any closed test edge in each preset closure direction, and calculating a second ratio of the number of gradient magnitude changes with positive signs to the number of gradient magnitude changes with negative signs; in response to the second ratio being 1, the region constructed by the closed test edge is a non-defect region; in response to the second ratio not being 1, the region constructed by the closed test edge is a real defect region.

[0008] Preferably, the calculation of the gradient magnitude change of any closed edge under test in any closed preset direction includes: constructing the minimum bounding rectangle of any closed edge under test, taking the intersection of the diagonals of the minimum bounding rectangle as the first center point of any closed edge under test, and taking the 8-neighborhood direction of the first center point as the closed preset direction; obtaining the gradient magnitude of each pixel of the first center point in any closed preset direction to construct an magnitude sequence, calculating the first-order difference sequence of the magnitude sequence, and summing the difference values ​​in the first-order difference sequence as the gradient magnitude change of any closed edge under test in any closed preset direction.

[0009] Preferably, the determination result of the non-closed test edge includes: growing any non-closed test edge according to a preset growth rule to obtain a fitted edge, and obtaining several fitted lines, and obtaining the midpoint of any fitted line; constructing the minimum bounding rectangle of the fitted edge, and taking the intersection of the diagonals of the minimum bounding rectangle as the second center point of the fitted edge; obtaining the direction of the line connecting the second center point and the midpoint of any fitted line, traversing to obtain all connecting directions, and taking the 8-neighborhood direction of the second center point and all connecting directions as the non-closed preset direction; calculating any non-closed preset direction of the non-closed test edge. Let the slope sequence be normalized in the direction, where the slope value in the slope sequence is the slope between two adjacent pixels in any non-closed preset direction; calculate the correlation between the slope sequence of any connecting direction and the slope sequence of any 8-neighboring direction, traverse to obtain the correlation between any connecting direction and each 8-neighboring direction, calculate the mean of the correlation between any connecting direction and all 8-neighboring directions, and in response to the existence of any mean correlation greater than a preset threshold, the region constructed by the non-closed test edge is a real defect region, and in response to the fact that the mean of all correlations is not greater than the preset threshold, the region constructed by the non-closed test edge is a non-defect region.

[0010] Preferably, the preset growth rule includes: taking the endpoint of the non-closed edge to be tested as the growth starting point, taking any initial growth point as the target point, and obtaining the neighboring pixels of the target point in the 8-neighborhood; taking the edge point adjacent to the target point as the first reference point, and taking the edge point adjacent to the first reference point (excluding the target point) as the second reference point; for any neighboring pixel, taking the angle between the line connecting the target point and the neighboring pixel and the line connecting the target point and the first reference point as the first angle, and taking the angle between the line connecting the first reference point and the second reference point and the line connecting the first reference point and the target point as the second angle, wherein both the first angle and the second angle are not greater than a straight angle; calculating the growth evaluation of the target point and any neighboring pixel, and the growth evaluation satisfies the following relationship: , Indicates the target point and neighboring pixels Growth evaluation, Indicates the target point grayscale value, Represents neighboring pixels grayscale value, Indicates the first angle. Indicates the second angle. It indicates a straight angle.

[0011] In a second aspect, an asphalt paving smoothness analysis system includes a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the asphalt paving smoothness analysis method described in any one of the claims is implemented.

[0012] The present invention has the following effects: This invention captures and analyzes high-definition images of paved asphalt pavements, combining edge detection and gradient amplitude variation calculation to effectively distinguish between real and non-defect areas, thus achieving high-precision smoothness analysis. By using edge evaluation and gradient amplitude variation calculation, defect detection becomes more accurate, identifying potential defects in the pavement and distinguishing them from non-defect areas. This technology provides real-time and accurate quality assessment for pavement construction, effectively improving the quality of asphalt paving, ensuring pavement smoothness meets standard requirements, and reducing subsequent maintenance and repair costs. Attached Figure Description

[0013] Figure 1 This is a flowchart of a method for analyzing the smoothness of asphalt paving according to an embodiment of the present invention. Detailed Implementation

[0014] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0015] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0016] Reference Figure 1 A method for analyzing the smoothness of asphalt paving includes steps S1-S2, as detailed below: S1: Obtain the surface grayscale image of the asphalt paved road after preprocessing, and perform edge detection on the surface grayscale image to obtain the edges.

[0017] In one embodiment, a camera installed at the rear can capture high-definition images of the paved road surface in real time during asphalt paving to ensure paving quality.

[0018] For example, the camera should be positioned 3 meters behind the paver, at a height of 1.5 meters above the ground, and mounted at a 30-degree downward angle to cover a road surface at least 3 meters wide. This avoids interference from other sources (such as other equipment or obstructions) that could affect image quality, ensuring the image accurately reflects the road surface condition. For accurate texture recognition, the camera needs a 1920×1080 resolution and the ability to clearly identify details down to 1mm. The camera's frame rate should be 15–25fps, equipped with a gimbal for image stabilization and an anti-fog / dust-proof lens to ensure stable shooting in various environments. To match the paver's operating speed, the camera's shooting frequency needs to be dynamically adjusted: decreasing to 5Hz when the paving speed is below 2m / min, and increasing to 10Hz when the speed exceeds 4m / min. This ensures high-quality road surface images are obtained at different paving speeds.

[0019] Once the road surface image is acquired, image preprocessing is required to eliminate various interference factors, such as dust and changes in lighting, to improve the accuracy of image processing. First, the color image is converted to a single-channel grayscale image, and the grayscale value of each pixel is extracted. To remove noise while preserving the boundary information of road surface bumps and depressions, a 3×3 window median filtering algorithm can be applied. Next, considering the often unstable lighting conditions at the paving site, an adaptive histogram equalization algorithm can be used. This not only effectively enhances image contrast but also better preserves image details, thereby improving the accuracy of subsequent smoothness analysis.

[0020] At this point, a grayscale image of the pre-treated asphalt pavement surface is obtained.

[0021] Edge detection is performed on the surface grayscale image using the Sobel operator to obtain the edges.

[0022] S2: Calculate the edge evaluation of any edge, retain the edges whose edge evaluation is greater than the preset evaluation threshold as the edge to be tested, determine the judgment result of any edge to be tested, and the judgment result is the real defect area and non-defect area. Complete the flatness analysis based on the judgment result.

[0023] In one embodiment, calculating the edge evaluation of any edge includes: taking pixels on the edge as edge points, converting the gradient direction of any edge point into a two-dimensional vector, calculating the mean of the two-dimensional vectors of all edge points in the same dimension, constructing a mean vector, and restoring the mean vector to the average angle using the arctangent function; taking the absolute difference between the gradient direction of any edge point and the average angle as the first difference, taking the difference between the full angle and the first difference as the second difference, selecting the minimum value of the first difference and the second difference of any edge point as the evaluation value, calculating the mean of the evaluation values ​​of all edge points, calculating the first ratio of 1 to the right angle, and calculating the product of the first ratio and the mean of the evaluation values, and taking the difference between 1 and the product as the edge evaluation.

[0024] It needs to be explained that converting the gradient direction of any edge point into a two-dimensional vector involves calculating the sine and cosine values ​​of the gradient direction, averaging the sine values ​​of all edge points, and averaging the cosine values ​​of all edge points to obtain the mean vector. The arctangent function is then used to convert the mean vector into an angle.

[0025] Marginal evaluation satisfies the following relation: , Indicates marginal evaluation, Represents edge points gradient direction, Indicates the average angle. Indicates the number of edge points. Indicates the first difference. Indicates the second worst. This indicates selecting the minimum value between the first and second differences.

[0026] In image processing, the gradient direction of an edge describes the direction of the fastest change in grayscale value and is commonly used to characterize the edge's orientation. For the edge of a closed region, the gradient direction of each pixel reflects the direction of the fastest change in grayscale value at that point, and this direction is perpendicular to the actual edge orientation. For example, if the grayscale value of an edge changes most drastically at that point, the gradient direction will point in the direction of the fastest increase in grayscale value, which is usually perpendicular to the tangent direction of the edge. For ease of description and consistency, the horizontal rightward direction of the image is usually chosen as the zero-degree reference, and the gradient direction is defined to increase counterclockwise. That is, the gradient direction of each pixel in the image is quantized into an angle, with the unit of angle being degrees, and it increases counterclockwise from the horizontal right side of the image.

[0027] Indicates the calculation of edge points The minimum loop angle difference between the gradient direction and the main direction is ensured to ensure that the angle difference does not exceed a horizontal angle, thus avoiding misjudgment of direction.

[0028] It represents the average minimum angle difference between the gradient direction of all edge points and the main direction, eliminating the deviation caused by noise in the angle difference of individual edge points, and reflecting the overall trend.

[0029] It can serve a normalization function, that is, normalize the deviation size, and in practical application scenarios The maximum is 90 degrees.

[0030] Real defects typically possess unique and stable physical structural characteristics, resulting in a certain regularity in the gradient direction of their edges, which usually remains consistent within a certain range. For example, the edges of defects are usually continuous and smooth, with minimal changes in gradient direction. In contrast, non-defect regions or foreign objects often have more complex and chaotic physical structures, leading to significant variations in the gradient direction at their edges and a lack of obvious regularity.

[0031] In actual testing, a preset evaluation threshold is set to distinguish between real defects and foreign objects. When the gradient direction change of a certain edge is greater than the preset threshold, it indicates that the edge may belong to a real defect, and its edge has strong regularity, so it is retained as the edge to be tested; while those edges whose gradient direction change is less than the threshold usually match the structural characteristics of foreign objects, because the edges of foreign objects are more disordered and the gradient direction cannot show consistency, and thus they are judged as foreign objects.

[0032] To determine the judgment result of any edge to be measured, where the edge to be measured includes closed edges and non-closed edges, it is necessary to determine the judgment result of closed edges and the judgment result of non-closed edges separately.

[0033] The determination results for closing the edge to be measured include: Calculate the gradient magnitude change of any closed edge under test in any preset closed direction, iterate through the edges to obtain the gradient magnitude change of any closed edge under test in each preset closed direction, and calculate a second ratio of the number of positive gradient magnitude changes to the number of negative gradient magnitude changes; if the second ratio is 1, the region constructed by the closed edge under test is a non-defect region; if the second ratio is not 1, the region constructed by the closed edge under test is a real defect region.

[0034] The calculation of the gradient magnitude change of any closed edge under test in any closed preset direction includes: constructing the minimum bounding rectangle of any closed edge under test, taking the intersection of the diagonals of the minimum bounding rectangle as the first center point of any closed edge under test, and taking the 8-neighborhood direction of the first center point as the closed preset direction; obtaining the gradient magnitude of each pixel of the first center point in any closed preset direction to construct an magnitude sequence, calculating the first difference sequence of the magnitude sequence, and summing the difference values ​​in the first difference sequence as the gradient magnitude change of any closed edge under test in any closed preset direction.

[0035] The determination results for identifying non-closed test edges include: According to the preset growth rules, any non-closed edge to be tested is grown to obtain a fitted edge, and several fitted lines are obtained. The midpoint of any fitted line is then obtained.

[0036] The preset growth rules include: Using the endpoints of the non-closed edge to be tested as the growth start point, any initial growth point is taken as the target point, and the neighboring pixels of the target point in the 8-neighborhood are obtained; the edge points adjacent to the target point are taken as the first reference points, and the edge points adjacent to the first reference points (excluding the target point) are taken as the second reference points; for any neighboring pixel, the angle between the line connecting the target point and the neighboring pixel and the line connecting the target point and the first reference point is taken as the first angle, and the angle between the line connecting the first reference point and the second reference point and the line connecting the first reference point and the target point is taken as the second angle, wherein the first angle and the second angle are both no greater than a straight angle.

[0037] Calculate the growth evaluation of the target point and any neighboring pixels. The growth evaluation satisfies the following relationship: , Indicates the target point and neighboring pixels Growth evaluation, Indicates the target point grayscale value, Represents neighboring pixels grayscale value, Indicates the first angle. Indicates the second angle. It indicates a straight angle.

[0038] It should be noted that all endpoints grow simultaneously until an intersection is formed and the non-closed edge to be tested becomes closed, at which point growth stops.

[0039] According to the preset growth rules, any non-closed edge to be tested is grown to obtain a fitted edge, and several fitted lines are obtained. The midpoint of any fitted line is obtained. Construct the minimum bounding rectangle of the fitted edge, and take the intersection of the diagonals of the minimum bounding rectangle as the second center point of the fitted edge; obtain the direction of the line connecting the second center point and the midpoint of any fitted line, starting from the second center point, and traverse to obtain all the line directions. Take the 8-neighborhood direction of the second center point and all the line directions as the non-closed preset direction.

[0040] It should be noted that since there may be more than one fitted edge in a non-closed test edge, the preset direction of the non-closed test edge will also change with the number of fitted edges.

[0041] Calculate the normalized slope sequence of any non-closed preset direction of the non-closed edge to be tested. The slope value in the slope sequence is the slope between two adjacent pixels in any non-closed preset direction. Calculate the correlation between the slope sequence of any connecting direction and the slope sequence of any 8-neighboring direction. Iterate to obtain the correlation between any connecting direction and each 8-neighboring direction. Calculate the mean correlation between any connecting direction and all 8-neighboring directions. If any mean correlation is greater than a preset threshold, the region constructed by the non-closed edge to be tested is a real defect region. If all mean correlations are not greater than the preset threshold, the region constructed by the non-closed edge to be tested is a non-defect region.

[0042] At this point, it can be determined whether all the edges to be tested in the surface grayscale image are real defect areas or non-defect areas. The ratio of the number of pixels in the real defect area to the number of pixels in the surface grayscale image is used as the flatness, thus completing the flatness analysis.

[0043] The system includes a processor and a memory, the memory storing computer program instructions, which, when executed by the processor, implement a method for analyzing the smoothness of asphalt paving according to the first aspect of the present invention.

[0044] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0045] It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept, and these all fall within the scope of protection of this invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A method for analyzing the smoothness of asphalt paving, characterized in that, include: Obtain the surface grayscale image of the pre-processed asphalt paved road surface, and perform edge detection on the surface grayscale image to obtain the edges; Calculate the edge evaluation of any edge, retain the edges whose edge evaluation is greater than the preset evaluation threshold as the edge to be tested, determine the judgment result of any edge to be tested, and the judgment result is the real defect area and non-defect area. Complete the flatness analysis based on the judgment result. The edge evaluation for any edge includes: Pixels on the edge are taken as edge points. The gradient direction of any edge point is converted into a two-dimensional vector. The mean of the two-dimensional vectors of all edge points in the same dimension is calculated to construct the mean vector. The mean vector is restored to the average angle by the arctangent function. The absolute difference between the gradient direction of any edge point and the average angle is taken as the first difference. The difference between the full angle and the first difference is taken as the second difference. The minimum value of the first difference and the second difference of any edge point is selected as the evaluation value. The mean of the evaluation values ​​of all edge points is calculated. The first ratio of 1 to the right angle is calculated. The product of the first ratio and the mean evaluation value is calculated. The difference between 1 and the product is taken as the edge evaluation.

2. The method for analyzing the smoothness of asphalt paving according to claim 1, characterized in that, The determination result for any edge to be measured includes: The edges to be measured include closed edges and non-closed edges; The judgment result for determining any edge to be measured includes the judgment result for determining a closed edge to be measured and the judgment result for determining a non-closed edge to be measured.

3. The method for analyzing the smoothness of asphalt paving according to claim 2, characterized in that, The determination result for identifying a closed edge to be measured includes: Calculate the gradient magnitude change of any closed edge under test in any closed preset direction, traverse to obtain the gradient magnitude change of any closed edge under test in each closed preset direction, and calculate the second ratio of the number of gradient magnitude changes with positive signs to the number of gradient magnitude changes with negative signs; When the second ratio is 1, the region formed by closing the edge to be tested is a non-defect region; when the second ratio is not 1, the region formed by closing the edge to be tested is a real defect region.

4. The method for analyzing the smoothness of asphalt paving according to claim 3, characterized in that, The calculation of the gradient magnitude change of any closed edge to be measured in any closed preset direction includes: Construct the minimum bounding rectangle of any closed edge to be tested, take the intersection of the diagonals of the minimum bounding rectangle as the first center point of any closed edge to be tested, and take the 8-neighborhood direction of the first center point as the closure preset direction; The gradient magnitude of each pixel at the first center point in any closed preset direction is obtained to construct an magnitude sequence. The first-order difference sequence of the magnitude sequence is calculated, and the difference values ​​in the first-order difference sequence are summed as the gradient magnitude change of any closed edge to be tested in any closed preset direction.

5. The method for analyzing the smoothness of asphalt paving according to claim 2, characterized in that, The determination results for identifying non-closed edges to be tested include: According to the preset growth rules, any non-closed edge to be tested is grown to obtain a fitted edge, and several fitted lines are obtained. The midpoint of any fitted line is obtained. Construct the minimum bounding rectangle of the fitted edge, and use the intersection of the diagonals of the minimum bounding rectangle as the second center point of the fitted edge; Obtain the line direction starting from the second center point and along the midpoint of the second center point and any fitted line. Traverse all line directions to obtain them. Use the 8 neighborhood directions of the second center point and all line directions as non-closed preset directions. Calculate the normalized slope sequence of any non-closed preset direction of the non-closed edge to be tested. The slope value in the slope sequence is the slope between two adjacent pixels in any non-closed preset direction. Calculate the correlation between the slope sequence of any connecting line direction and the slope sequence of any 8-neighboring direction. Iterate through the correlation between any connecting line direction and each 8-neighboring direction. Calculate the mean correlation between any connecting line direction and all 8-neighboring directions. If any mean correlation is greater than a preset threshold, the region constructed by the non-closed test edge is a real defect region. If all mean correlations are not greater than the preset threshold, the region constructed by the non-closed test edge is a non-defect region.

6. The method for analyzing the smoothness of asphalt paving according to claim 5, characterized in that, The preset growth rules include: Using the endpoints of the non-closed edge to be tested as the growth start point, and taking any initial growth point as the target point, the neighboring pixels of the target point in the 8-neighborhood are obtained. The edge points adjacent to the target point are used as the first reference points, and the edge points other than the target point that are adjacent to the first reference points are used as the second reference points. For any neighboring pixel, the angle between the line connecting the target point and the neighboring pixel and the line connecting the target point and the first reference point is taken as the first angle, and the angle between the line connecting the first reference point and the second reference point and the line connecting the first reference point and the target point is taken as the second angle, wherein the first angle and the second angle are both no greater than a straight angle. Calculate the growth evaluation of the target point and any neighboring pixels. The growth evaluation satisfies the following relationship: , Indicates the target point and neighboring pixels Growth evaluation, Indicates the target point grayscale value, Represents neighboring pixels grayscale value, Indicates the first angle. Indicates the second angle. It indicates a straight angle.

7. A system for analyzing the smoothness of asphalt paving, characterized in that, include: A processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement a method for analyzing the smoothness of asphalt paving according to any one of claims 1-6.

Images