Intelligent recognition method and system for road surface defects based on machine vision

Abstract

The present invention relates to the field of image processing technology, and specifically to a method and system for intelligently identifying road defects based on machine vision; obtaining local disorder according to grayscale difference characteristics within a preset neighborhood range of pixel points in a road grayscale image; obtaining an adaptive window of a pixel point within a preset neighborhood range according to the difference characteristics of the gradient values ​​of the pixel point in the horizontal and vertical directions and the local disorder; obtaining a growth index according to the shape difference characteristics of the adaptive window of the pixel point and the grayscale difference characteristics within the adaptive window between the pixel point and the preset associated pixel point; performing regional growth according to the growth index of the pixel point and the preset associated pixel point to obtain a road surface area. The present invention obtains a non-defect confidence according to the grayscale value distribution characteristics of the road surface area and the angle distribution characteristics of the adjacent edge pixels; determines the defective area of ​​the road surface according to the non-defect confidence, thereby improving the detection accuracy of the defective area.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a method and system for intelligently identifying road surface defects based on machine vision. Background Art

[0002] At present, common roads are mainly cement pavements and asphalt pavements. After long-term use, due to factors such as their own structural strength and settlement, cracks and potholes will appear on the road surface, which will affect the comfort and safety of vehicle driving. Therefore, roads need to be inspected and maintained frequently. Using machine vision to detect and identify road defects can reduce the impact on vehicles during detection, reduce detection costs and improve detection efficiency. Existing machine vision can detect and extract defective areas in road images through region growing algorithms, but due to the roughness of the road surface and the presence of some repaired roads, the error of region extraction using the grayscale value between adjacent pixels as the judgment condition for region growth is large, which makes the accuracy of segmentation of different road surface areas low; ultimately affecting the accuracy of recognition and detection of road surface defect areas. Summary of the invention

[0003] In order to solve the technical problem of low accuracy of road defect detection using the traditional region growing algorithm, the present invention aims to provide a method and system for intelligent recognition of road defects based on machine vision. The technical solutions adopted are as follows:

[0004] Obtaining a road grayscale image;

[0005] Obtaining the local disorder degree of the pixel point according to the grayscale difference characteristics within a preset neighborhood range of the pixel point in the road grayscale image; obtaining the adaptive window of the pixel point according to the difference characteristics of the gradient values ​​of the pixel point in the horizontal direction and the vertical direction and the local disorder degree within the preset neighborhood range;

[0006] A growth index is obtained according to the shape difference characteristics of the adaptive window between the pixel point and the preset associated pixel point and the grayscale difference characteristics within the adaptive window; regional growth is performed according to the growth index of the pixel point and the preset associated pixel point to obtain different road surface areas; a non-defect confidence is obtained according to the grayscale value distribution characteristics of the road surface area and the angle distribution characteristics of the adjacent edge pixel points;

[0007] The defective area of ​​the road surface is determined according to the non-defective confidence level of the road surface area.

[0008] Furthermore, the step of obtaining the local disorder degree of the pixel point according to the grayscale difference feature within a preset neighborhood range of the pixel point in the road grayscale image comprises:

[0009] The average value of the absolute value of the grayscale difference between any pixel point and other pixel points within the preset neighborhood range of the pixel point is calculated to obtain the grayscale difference characteristic value of the arbitrary pixel point; the average value of all grayscale difference characteristic values ​​within the preset neighborhood range is calculated to obtain the local chaos degree of the pixel point.

[0010] Furthermore, the step of obtaining the adaptive window of the pixel point within the preset neighborhood according to the difference characteristics of the gradient values ​​of the pixel point in the horizontal direction and the vertical direction and the local chaos degree includes:

[0011] Calculate the sum of the average gradient in the horizontal direction and the average gradient in the vertical direction of the pixel point within the preset neighborhood range to obtain a comprehensive gradient value; calculate the sum of the comprehensive gradient value and a preset minimum positive number to obtain a gradient characterization value; calculate the ratio of the average gradient in the vertical direction to the gradient characterization value to obtain a horizontal texture feature value; calculate the product of the horizontal texture feature value and the local chaos degree and normalize them to obtain a horizontal length coefficient; calculate the sum of the horizontal length coefficient and a constant 1 to obtain a horizontal length reference; calculate the product of the horizontal length reference and a preset length and round them to obtain the horizontal length of the pixel point;

[0012] Calculate the ratio of the horizontal gradient average value to the gradient characterization value to obtain the vertical texture eigenvalue; calculate the product of the vertical texture eigenvalue and the local chaos degree and normalize them to obtain the vertical length coefficient; calculate the sum of the vertical length coefficient and the constant 1 to obtain the vertical length reference; calculate the product of the vertical length reference and the preset length and round it to obtain the vertical length of the pixel point; take the pixel point as the window center, the horizontal length as the window length, and the vertical length as the window width to obtain the adaptive window of the pixel point.

[0013] Furthermore, the step of obtaining the growth index according to the shape difference characteristics of the adaptive window between the pixel point and the preset associated pixel point and the grayscale difference characteristics within the adaptive window comprises:

[0014] Calculate the ratio of the horizontal length to the vertical length of the adaptive window to obtain the shape feature value of the adaptive window; calculate the average value of the grayscale value within the adaptive window to obtain the grayscale feature value; calculate the absolute value of the difference between the shape feature value of the pixel point and the preset associated pixel point to obtain a first difference value; calculate the absolute value of the difference between the grayscale feature value of the pixel point and the preset associated pixel point to obtain a second difference value; calculate the product of the first difference value and the second difference value and negatively correlate them to obtain the growth index of the pixel point and the preset associated pixel point.

[0015] Furthermore, the step of performing regional growth according to the growth index of the pixel point and the preset associated pixel point to obtain different road surface areas includes:

[0016] When the growth index of the pixel point and the preset associated pixel point exceeds the preset growth threshold, the pixel point and the preset associated pixel point perform regional growth to become the same road surface area; when the growth index of the pixel point and the non-directly connected preset associated pixel point exceeds the preset threshold, the pixel point and the non-directly connected preset associated pixel point and the interval pixel point perform regional growth to become the same road surface area; the road surface grayscale image is traversed to obtain different road surface areas.

[0017] Furthermore, the step of obtaining the non-defect confidence level according to the grayscale value distribution characteristics of the road surface area and the angle distribution characteristics of adjacent edge pixels includes:

[0018] The standard deviation of the grayscale values ​​of all pixels in the pavement area is calculated to obtain a chaotic eigenvalue; the angle between any edge pixel and the line connecting other edge pixels on both sides is calculated to obtain an angle eigenvalue; the standard deviation of the angle eigenvalues ​​of all edge pixels in the pavement area is calculated to obtain an angle discrete value; the product of the chaotic eigenvalue and the angle discrete value is calculated and negatively correlated to obtain the non-defect confidence.

[0019] Furthermore, the step of determining the defective area of ​​the road surface according to the non-defect confidence of the road surface area includes:

[0020] When the non-defect confidence level is lower than a preset defect threshold, the road surface area is a defect area.

[0021] The present invention also proposes a road surface defect intelligent identification system based on machine vision, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement any step of a road surface defect intelligent identification method based on machine vision.

[0022] The present invention has the following beneficial effects:

[0023] In the present invention, obtaining the local disorder degree can reflect the amount of information near the pixel point, so that the adaptive window can more accurately characterize the local characteristics of the pixel point, and preliminarily improve the accuracy of regional growth; obtaining the adaptive window can adaptively adjust the window size and proportion according to the texture characteristics near the pixel point, so that the texture characteristics of the pixel can be taken into account when the subsequent regional growth is performed, and the accuracy of regional growth is further improved. Obtaining the growth index can characterize the degree of association between the pixels according to the shape difference characteristics and grayscale difference characteristics of the adaptive window, reduce the error of the existing regional growth algorithm that only performs regional growth based on grayscale characteristics due to the rough road surface and the presence of repairs, and improve the accuracy of regional growth. Obtaining different road surface areas according to the growth index can accurately segment different types of road surface areas and improve the detection accuracy of defective areas. Obtaining the non-defect confidence of the road surface area can accurately indicate the possibility of the existence of a defective area according to the characteristic difference between the defective area and the normal area, and improve the accuracy of road defect detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] 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.

[0025] Figure 1 A flow chart of a method for intelligently identifying road surface defects based on machine vision provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0026] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of a road surface defect intelligent identification method and system based on machine vision proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0027] 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.

[0028] The following is a detailed description of a specific solution of a method and system for intelligently identifying road defects based on machine vision provided by the present invention in conjunction with the accompanying drawings.

[0029] See also Figure 1, which shows a flow chart of a method for intelligently identifying road surface defects based on machine vision provided by an embodiment of the present invention, the method comprising the following steps:

[0030] Step S1, obtaining a road surface grayscale image.

[0031] In the embodiment of the present invention, the implementation scenario is to detect road defects based on machine vision to improve the accuracy of road defect detection. First, a road surface grayscale image is obtained, and a road surface range to be detected is photographed by a drone or a road surface detection vehicle to obtain a road surface grayscale image for defect area detection.

[0032] Step S2, obtaining the local chaos of the pixel point according to the grayscale difference characteristics within a preset neighborhood range of the pixel point in the road grayscale image; obtaining an adaptive window of the pixel point according to the difference characteristics of the gradient values ​​in the horizontal and vertical directions of the pixel point and the local chaos within the preset neighborhood range.

[0033] The traditional region growing algorithm determines whether it is the same region by the gray value difference between a certain pixel and its neighboring pixels; however, for the road surface area, due to its surface roughness, for example, the color of the gap between the gravel and the gravel may be inconsistent, and the color of the repaired local area is different from that of the normal road surface, but they may all belong to the normal road surface; therefore, it is difficult to accurately detect the defective area through the traditional region growing algorithm, and the judgment method of region growing needs to be improved. Since there are obvious differences between the defective area and the normal road surface area, the region growth can be performed by the similarity of the local features between the pixels to obtain different regions, and then it is necessary to first determine the local range of the pixel for region growth judgment. The greater the grayscale difference around the pixel, the more information the pixel area represents, and the larger the local range needs to be, thereby improving the accuracy of region growth judgment; therefore, the local disorder of the pixel is obtained according to the grayscale difference characteristics within the preset neighborhood range of the pixel in the road surface grayscale image.

[0034] Preferably, in an embodiment of the present invention, the step of obtaining the local chaos degree includes: calculating the average value of the absolute value of the grayscale difference between any pixel point and other pixel points within the preset neighborhood range of the pixel point, and obtaining the grayscale difference characteristic value of the arbitrary pixel point; in an embodiment of the present invention, the preset neighborhood range is a window range with a length of 3 pixels centered on the pixel point; the larger the grayscale difference characteristic value, the greater the grayscale difference between the arbitrary pixel point and other pixel points. Calculate the average value of all grayscale difference characteristic values ​​within the preset neighborhood range to obtain the local chaos degree of the pixel point; the larger the grayscale difference within the preset neighborhood range of the pixel point, the greater the local chaos degree, and the pixel point needs a larger local range to represent more information when performing regional growth judgment. The formula for obtaining the local chaos degree includes:

[0035]

[0036] In the formula, represents the local disorder degree of the i-th pixel, M represents the number of pixels within the preset neighborhood, represents the gray value of the mth pixel in the preset neighborhood of the i-th pixel. represents the gray value of the rth pixel except the mth pixel within the preset neighborhood of the i-th pixel. Represents the grayscale difference eigenvalue of the m-th pixel.

[0037] Furthermore, when there is a local texture change in a pixel point, it is necessary to appropriately expand the window in the direction of texture extension. For example, when two pixels have horizontal textures in the preset neighborhood, the horizontal ratio of the local window of the two pixels should be larger; even if the local colors of the two pixels are different, the textures in the corresponding local windows have certain similarities and may belong to the same road surface area, effectively avoiding the segmentation of the repaired road surface into multiple road surface areas due to the different colors of the normal road surface, which affects the accuracy of road surface defect detection. Therefore, the adaptive window of the pixel point can be obtained in the preset neighborhood according to the difference characteristics of the gradient values ​​of the horizontal and vertical directions of the pixel point and the local chaos.

[0038] Preferably, in an embodiment of the present invention, the step of obtaining an adaptive window includes: calculating the sum of the average value of the gradient in the horizontal direction and the average value of the gradient in the vertical direction of the pixel within a preset neighborhood range to obtain a comprehensive gradient value; calculating the sum of the comprehensive gradient value and a preset minimum positive number to obtain a gradient representation value; in an embodiment of the present invention, the preset minimum positive number is 0.01, the purpose is to avoid the situation where the denominator is 0. Calculate the ratio of the average value of the gradient in the vertical direction to the gradient representation value to obtain a horizontal texture feature value; when the horizontal texture feature value is larger, it means that the average value of the gradient in the horizontal direction is smaller, the proportion of the average value of the gradient in the vertical direction in the gradient representation value is larger, and the local texture of the pixel point is stronger in the horizontal direction, and the horizontal length of the window should be extended. Calculate the product of the horizontal texture feature value and the local chaos degree and normalize them to obtain a horizontal length coefficient; when the horizontal texture feature value and the local chaos degree are larger, the horizontal length coefficient is larger, which means that the window size of the pixel point in the horizontal direction is larger. Calculate the sum of the horizontal length coefficient and the constant 1 to obtain the horizontal length reference; calculate the product of the horizontal length reference and the preset length and round it to obtain the horizontal length of the pixel; in the embodiment of the present invention, the preset length is 5, which can be determined by the implementer according to the implementation scenario. The preset length needs to exceed the side length of the preset neighborhood range and needs to be rounded to an odd number. The greater the grayscale difference of the local pixel point and the more the texture is oriented to the horizontal direction, the longer the horizontal length of the local window of the pixel point.

[0039] Similarly, calculate the ratio of the horizontal gradient average to the gradient characterization value to obtain the vertical texture eigenvalue; the larger the vertical texture eigenvalue, the stronger the local texture of the pixel in the vertical direction. Calculate the product of the vertical texture eigenvalue and the local chaos degree and normalize them to obtain the vertical length coefficient; the larger the vertical length coefficient, the larger the vertical window size of the pixel. Calculate the sum of the vertical length coefficient and the constant 1 to obtain the vertical length reference; calculate the product of the vertical length reference and the preset length and round it to obtain the vertical length of the pixel. Then use the pixel as the window center, the horizontal length as the window length, and the vertical length as the window width to obtain the adaptive window of the pixel. The formula for obtaining the vertical length includes:

[0040]

[0041] In the formula, Represents the vertical length of the adaptive window of the i-th pixel, Indicates rounding the function, T indicates the preset length, represents normalization, represents the local disorder of the ith pixel, a represents a preset minimum positive number, K represents the average gradient value in the horizontal direction, G represents the comprehensive gradient value, represents the gradient characterization value, represents the vertical texture eigenvalue, represents the vertical length coefficient, Indicates the vertical length reference.

[0042] Step S3, obtaining a growth index based on the shape difference characteristics of the adaptive window between the pixel point and the preset associated pixel point and the grayscale difference characteristics within the adaptive window; performing regional growth based on the growth index of the pixel point and the preset associated pixel point to obtain different road surface areas; obtaining a non-defect confidence based on the grayscale value distribution characteristics of the road surface area and the angle distribution characteristics of adjacent edge pixels.

[0043] After obtaining the adaptive window of each pixel point of the road grayscale image, regional growth can be performed according to the characteristic difference of the adaptive window between the pixels to obtain different road surface areas; therefore, the growth index is obtained according to the shape difference characteristics of the adaptive window between the pixel point and the preset associated pixel point and the grayscale difference characteristics within the adaptive window; preferably, in the embodiment of the present invention, the step of obtaining the growth index includes: calculating the ratio of the horizontal length to the vertical length of the adaptive window to obtain the shape feature value of the adaptive window; since the values ​​of the horizontal length and the vertical length of the adaptive window are related to the texture direction near the pixel point, the value of the shape feature value can characterize the texture direction near the pixel point. Calculate the average value of the grayscale value within the adaptive window to obtain the grayscale feature value; the grayscale feature value can characterize the grayscale feature near the pixel point. Calculate the absolute value of the difference between the shape feature value of the pixel point and the preset associated pixel point to obtain the first difference value; when the first difference value is smaller, it means that the texture directions near the two pixels are more likely to be the same, and the pixels are more likely to be in the same road surface area, which reduces the probability of being divided into different road surface areas due to the difference in grayscale between the partially repaired road surface and the normal road surface. Calculate the absolute value of the difference between the grayscale feature value of the pixel point and the preset associated pixel point to obtain a second difference value; when the second difference value is smaller, it means that the grayscale features near the two pixels are more similar, and the more likely they are pixels in the same road area. In an embodiment of the present invention, the preset associated pixel point is a pixel point within a range with the pixel point as the center and a side length of 5 pixels. Calculate the product of the first difference value and the second difference value and negatively correlate them to obtain the growth index of the pixel point and the preset associated pixel point; when the first difference value and the second difference value are smaller, the growth index is larger, and the pixel point and the preset associated pixel point are more likely to be pixels in the same area.

[0044] Further, after the growth index of the pixel point and the preset associated pixel point is obtained, regional growth can be performed according to the growth index of the pixel point and the preset associated pixel point to obtain different road surface areas; preferably, in the embodiment of the present invention, the step of obtaining the road surface area includes: when the growth index of the pixel point and the preset associated pixel point exceeds the preset growth threshold, the pixel point and the preset associated pixel point perform regional growth to become the same road surface area; in the embodiment of the present invention, the preset growth threshold is 0.6, which can be determined by the implementer according to the implementation scenario. When the growth index of the pixel point and the preset associated pixel point that is not directly connected exceeds the preset threshold, the pixel point and the preset associated pixel point that is not directly connected and the interval pixel point perform regional growth to become the same road surface area; for example, if the growth index of the pixel point and the left adjacent pixel point is less than the preset growth threshold, but the growth index of the pixel point and the left interval pixel point exceeds the preset growth threshold, then the pixel point and the left pixel point and the interval pixel point are the same road surface area; thereby reducing the situation of segmenting the same road surface area due to road roughness and road surface fine lines. The road surface grayscale image is traversed to obtain different road surface areas, which preliminarily improves the detection accuracy of defective areas.

[0045] After obtaining different road surface areas, if there is only one road surface area in the entire road surface grayscale image, it means that the road in this area is normal and there is no defective area with road damage; if there are different road surface areas, it means that there may be road marking areas, smeared areas and defective areas, which need to be further distinguished. For defective areas, the defect edges are irregular, and the grayscale values ​​of the pixels in the area are relatively discrete; therefore, the non-defect confidence can be obtained based on the distribution characteristics of the grayscale values ​​of the edge pixels of the road surface area and the angular distribution characteristics of the adjacent edge pixels.

[0046] Preferably, in an embodiment of the present invention, the step of obtaining the non-defect confidence includes: calculating the standard deviation of the grayscale values ​​of all pixels in the road surface area to obtain a chaotic eigenvalue; when the chaotic eigenvalue is larger, it means that the grayscale values ​​of the pixels in the road surface area are more discrete, and it is more likely to be a defective area with damaged roads. Calculate the angle between any edge pixel point and the line connecting other edge pixel points on both sides to obtain an angle eigenvalue; calculate the standard deviation of the angle eigenvalues ​​of all edge pixels in the road surface area to obtain an angle discrete value; when the angle discrete value is larger, it means that the edge of the road surface area is more irregular, and it is more likely to be a defective area with damaged roads. Calculate the product of the chaotic eigenvalue and the angle discrete value and negatively correlate them to obtain the non-defect confidence; when the chaotic eigenvalue and the angle discrete value are larger, the non-defect confidence is smaller, and the road surface area is more likely to be a defective area.

[0047] Step S4, determining the defective area of ​​the road surface according to the non-defective confidence level of the road surface area.

[0048] After obtaining the non-defect confidence of all road surface areas, the defective area of ​​the road surface can be determined according to the non-defect confidence of the road surface area, specifically including: when the non-defect confidence is lower than the preset defect threshold, the road surface area is a defective area. In the embodiment of the present invention, the preset defect threshold is 0.5, which can be determined by the implementer according to the implementation scenario. Then, road maintenance can be performed according to the specific location of the defective area. So far, an adaptive window is obtained according to the grayscale features and texture features near the pixel point, and regional growth is performed according to the size and grayscale features of the adaptive window, which improves the accuracy of regional growth in rough road surfaces and road repairs, makes the segmentation of different road surface areas more accurate, and finally improves the accuracy of road surface defect detection.

[0049] In summary, the embodiment of the present invention provides a method for intelligently identifying road defects based on machine vision; local chaos is obtained according to the grayscale difference characteristics within a preset neighborhood range of pixels in a road grayscale image; an adaptive window of pixels is obtained according to the difference characteristics of the gradient values ​​in the horizontal and vertical directions of pixels and the local chaos within the preset neighborhood range; a growth index is obtained according to the shape difference characteristics of the adaptive window of pixels and preset associated pixels and the grayscale difference characteristics within the adaptive window; regional growth is performed according to the growth index of pixels and preset associated pixels to obtain a road surface area. The present invention obtains a non-defect confidence according to the grayscale value distribution characteristics of the road surface area and the angle distribution characteristics of adjacent edge pixels; the defect area of ​​the road surface is determined according to the non-defect confidence, thereby improving the detection accuracy of the defect area.

[0050] The present invention also proposes a road surface defect intelligent identification system based on machine vision, including a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the computer program to implement any step of a road surface defect intelligent identification method based on machine vision.

[0051] 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.

[0052] 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.

Claims

1. A method for intelligently identifying road surface defects based on machine vision, characterized in that: The method comprises the following steps: Obtaining a road grayscale image; Obtaining the local disorder degree of the pixel point according to the grayscale difference characteristics within a preset neighborhood range of the pixel point in the road grayscale image; obtaining the adaptive window of the pixel point according to the difference characteristics of the gradient values ​​of the pixel point in the horizontal direction and the vertical direction and the local disorder degree within the preset neighborhood range; A growth index is obtained according to the shape difference characteristics of the adaptive window between the pixel point and the preset associated pixel point and the grayscale difference characteristics within the adaptive window; regional growth is performed according to the growth index of the pixel point and the preset associated pixel point to obtain different road surface areas; a non-defect confidence is obtained according to the grayscale value distribution characteristics of the road surface area and the angle distribution characteristics of the adjacent edge pixel points; Determining a defective area of ​​the road surface according to a non-defective confidence level of the road surface area; The step of obtaining the adaptive window of the pixel point according to the difference characteristics of the gradient values ​​of the pixel point in the horizontal direction and the vertical direction and the local chaos within the preset neighborhood range comprises: Calculate the sum of the average gradient in the horizontal direction and the average gradient in the vertical direction of the pixel point within the preset neighborhood range to obtain a comprehensive gradient value; calculate the sum of the comprehensive gradient value and a preset minimum positive number to obtain a gradient characterization value; calculate the ratio of the average gradient in the vertical direction to the gradient characterization value to obtain a horizontal texture feature value; calculate the product of the horizontal texture feature value and the local chaos degree and normalize them to obtain a horizontal length coefficient; calculate the sum of the horizontal length coefficient and a constant 1 to obtain a horizontal length reference; calculate the product of the horizontal length reference and a preset length and round them to obtain the horizontal length of the pixel point; Calculate the ratio of the average value of the gradient in the horizontal direction to the gradient characterization value to obtain a vertical texture feature value; calculate the product of the vertical texture feature value and the local chaos degree and normalize them to obtain a vertical length coefficient; calculate the sum of the vertical length coefficient and a constant 1 to obtain a vertical length reference; calculate the product of the vertical length reference and a preset length and round it to obtain the vertical length of the pixel point; take the pixel point as the window center, the horizontal length as the window length, and the vertical length as the window width to obtain an adaptive window for the pixel point; The step of performing regional growth according to the growth index of the pixel point and the preset associated pixel point to obtain different road surface areas comprises: When the growth index of the pixel point and the preset associated pixel point exceeds the preset growth threshold, the pixel point and the preset associated pixel point perform regional growth to become the same road surface area; the preset associated pixel point is any other pixel point within the range of 5 other pixel points with the pixel point as the center; when the growth index of the pixel point and the non-directly connected preset associated pixel point exceeds the preset threshold, the pixel point and the non-directly connected preset associated pixel point and the interval pixel point perform regional growth to become the same road surface area; the road surface grayscale image is traversed to obtain different road surface areas.

2. The method for intelligently identifying road surface defects based on machine vision according to claim 1, characterized in that: The step of obtaining the local disorder degree of the pixel point according to the grayscale difference characteristics within a preset neighborhood range of the pixel point in the road grayscale image comprises: The average value of the absolute value of the grayscale difference between any pixel point and other pixel points within the preset neighborhood range of the pixel point is calculated to obtain the grayscale difference characteristic value of the arbitrary pixel point; the average value of all grayscale difference characteristic values ​​within the preset neighborhood range is calculated to obtain the local chaos degree of the pixel point.

3. The method for intelligently identifying road defects based on machine vision according to claim 1, characterized in that: The step of obtaining the growth index according to the shape difference characteristics of the adaptive window between the pixel point and the preset associated pixel point and the grayscale difference characteristics within the adaptive window comprises: Calculate the ratio of the horizontal length to the vertical length of the adaptive window to obtain the shape feature value of the adaptive window; calculate the average value of the grayscale value within the adaptive window to obtain the grayscale feature value; calculate the absolute value of the difference between the shape feature value of the pixel point and the preset associated pixel point to obtain a first difference value; calculate the absolute value of the difference between the grayscale feature value of the pixel point and the preset associated pixel point to obtain a second difference value; calculate the product of the first difference value and the second difference value and negatively correlate them to obtain the growth index of the pixel point and the preset associated pixel point.

4. The method for intelligently identifying road surface defects based on machine vision according to claim 1, characterized in that: The step of obtaining the non-defect confidence according to the gray value distribution characteristics of the road surface area and the angle distribution characteristics of adjacent edge pixels comprises: The standard deviation of the grayscale values ​​of all pixels in the pavement area is calculated to obtain a chaotic eigenvalue; the angle between any edge pixel and the line connecting other edge pixels on both sides is calculated to obtain an angle eigenvalue; the standard deviation of the angle eigenvalues ​​of all edge pixels in the pavement area is calculated to obtain an angle discrete value; the product of the chaotic eigenvalue and the angle discrete value is calculated and negatively correlated to obtain the non-defect confidence.

5. The method for intelligently identifying road defects based on machine vision according to claim 1, characterized in that: The step of determining the defective area of ​​the road surface according to the non-defect confidence of the road surface area comprises: When the non-defect confidence level is lower than a preset defect threshold, the road surface area is a defect area.

6. A road surface defect intelligent recognition system based on machine vision, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 5.

Images