A machine vision-based road surface disease identification method

Abstract

The present application relates to the technical field of road surface detection, and particularly relates to a road surface disease identification method based on machine vision; position registration is performed according to the gray difference features of left and right road surface images; an illumination response value is obtained according to the local gray distribution features of any position in the left and right road surface images; an illumination response difference value is obtained according to the difference features of the illumination response values under different illumination conditions and the standard illumination response difference features; an optimal reverse response difference value is obtained according to the difference features of the illumination response difference values on both sides of any position in any angle direction; and a local recess response point and a disease extension direction are obtained according to the data distribution features of the optimal reverse response difference values of all positions. The present application obtains a disease opening path according to the position distribution features of the local recess response point and the disease extension direction; and road disease marking is performed according to the disease opening path, thereby improving the accuracy of disease identification.

Description

Technical Field

[0001] This invention relates to the field of road surface inspection technology, and more specifically to a method for identifying road surface defects based on machine vision. Background Technology

[0002] Road surfaces are prone to various defects such as single cracks, network cracks, potholes, and spalling under the influence of vehicle loads, temperature changes, moisture erosion, and construction quality. If these defects are not detected and addressed promptly, they will accelerate the deterioration of the road surface and affect vehicle safety. Currently, road surface defect detection typically uses vehicle-mounted cameras to capture images of the road surface, then identifies defect areas through threshold segmentation, edge detection, texture analysis, or deep learning models. These methods usually rely on grayscale anomalies, texture anomalies, or shape anomalies in the road surface images for identification. However, during actual road inspections, tree shadows, vehicle shadows, oil stains, tire tracks, road patches, and water accumulation can also create localized dark lines, bright spots, or edge variations in the images. These interfering elements may appear similar to cracks or potholes in a single image, thus affecting the accuracy of road surface defect identification. Summary of the Invention

[0003] To address the aforementioned technical problems, the present invention aims to provide a method for identifying road surface defects based on machine vision, the specific technical solution of which is as follows: Install two lighting devices, one on the left and one on the right, onto the inspection vehicle to obtain images of the road surface under left-side lighting conditions and images of the road surface under right-side lighting conditions. Position registration is performed based on the grayscale difference characteristics of the overlapping area of ​​the left and right road surface images; the illumination response value of any location under different lighting conditions is obtained based on the local grayscale distribution characteristics of any location in the left and right road surface images; the illumination response difference value is obtained based on the difference characteristics of the illumination response values ​​of any location under different lighting conditions and the standard illumination response difference characteristics. The reverse response difference value is obtained based on the difference characteristics of the lighting response difference values ​​on both sides of any position at any angle; the optimal reverse response difference value is obtained based on the difference characteristics of the reverse response difference values ​​at different angles of the same position; the local depression response point and the corresponding disease extension direction are obtained based on the data distribution characteristics of the optimal reverse response difference values ​​at all positions; and the disease opening path is obtained based on the positional distribution characteristics of the local depression response point and the disease extension direction. Road defects are marked according to the opening paths of the defects.

[0004] Further, the step of performing position registration based on the grayscale difference features of the overlapping area of ​​the left-facing road surface image and the right-facing road surface image includes: In the road plane coordinate system, the right-side road surface image is fixed, and the left-side road surface image is moved. In the formula... The lateral displacement of the left-facing road surface image is... Longitudinal displacement is The image's overall grayscale difference value at that time; This represents the set of reference locations in the overlapping area of ​​the left and right road surface images. Let N represent any reference position in the set of reference positions, and let N represent the number of positions in the set of reference positions. Indicates the reference position in the right-hand road surface image. grayscale value, The lateral displacement of the left-facing road surface image is... Longitudinal displacement is Time reference position The grayscale value; the lateral displacement corresponding to the minimum value of the overall grayscale difference of the image is... and longitudinal displacement are Move the left-facing road surface image and then perform positional registration between the right-facing road surface image and the moved left-facing road surface image.

[0005] Further, the step of obtaining the illumination response value of any location under different lighting conditions based on the local grayscale distribution features of any location in the left-illuminated road surface image and the right-illuminated road surface image includes: In the formula, k represents the lighting sign; when k is L, it represents the image of the road surface illuminated from the left; when k is R, it represents the image of the road surface illuminated from the right. This represents the lighting response value at any location i in the road surface image when the lighting identifier is k. This represents the grayscale value of any position i in the road surface image when the lighting indicator is k. This represents the median gray level of a preset neighborhood of any position i in the road surface image when the lighting identifier is k. This represents the absolute median deviation of the preset neighborhood of any position i in the road surface image when the lighting label is k. This indicates a preset minimum positive number.

[0006] Further, the step of obtaining the lighting response difference value based on the difference characteristics of the lighting response values ​​at any location under different lighting conditions and the standard lighting response difference characteristics includes: In the formula This represents the difference in lighting response at any location i. This represents the illumination response value at any position i in the left-facing road surface image. This represents the illumination response value at any position i in the right-side road surface image. This represents the difference in illumination response values ​​at any position i between the left-facing and right-facing road surface images corresponding to the preset standard road surface.

[0007] Furthermore, the step of obtaining the reverse response difference value based on the difference characteristics of the illumination response difference values ​​on both sides at any position in any angular direction includes: In the formula, Q represents the difference in the reverse response at any position in any angular direction, and G represents the set of indices of adjacent positions on both sides of the arbitrary position in the arbitrary angular direction. This represents the difference in illumination response at the g-th position, taken as a starting point, at any end of the arbitrary angular direction. This represents the difference in illumination response at the g-th position at the other end of the arbitrary angular direction. This indicates a preset minimum positive number.

[0008] Further, the step of obtaining the optimal reverse response difference value based on the difference characteristics of the reverse response difference values ​​at different angular directions at the arbitrary position includes: The maximum value of the difference in the reverse response at any given position in different angular directions is taken as the optimal difference in the reverse response at that given position.

[0009] Furthermore, the step of obtaining the local depression response point and the corresponding disease extension direction based on the data distribution characteristics of the best reverse response difference values ​​at all locations includes: The location where the optimal reverse response difference value exceeds a preset difference threshold is taken as the local depression response point; the angle direction in which the optimal reverse response difference value of the local depression response point is obtained is taken as the target direction, and the vertical direction of the target direction is taken as the disease extension direction.

[0010] Furthermore, the step of obtaining the opening path of the lesion based on the location distribution characteristics of the local depression response points and the lesion extension direction includes: In the pathological extension direction of the local depression response point, other connected local depression response points are found with a preset search step size and connected. By traversing all local depression response points, different pathological opening paths are obtained.

[0011] The present invention has the following beneficial effects: In this invention, position registration is performed based on the grayscale difference characteristics of the overlapping area between the left and right road surface images. This ensures that the same location in both images represents the same road surface entity, reducing subsequent analysis errors in lighting response and initially improving the accuracy of defect identification. Obtaining lighting response values ​​characterizes the degree of brightness difference between any given location and the local background; obtaining lighting response difference values ​​determines the lighting response differences of any given location under different directional lighting conditions. After the lighting direction changes, the light reception relationship on both sides of the depression or crack structure shifts. Therefore, obtaining the reverse response difference value can determine the difference in lighting response differences on both sides of any given location at different angles; furthermore, obtaining the optimal reverse response difference value can measure the probability of defects existing at that given location. Obtaining local depression response points and defect extension directions determines the distribution of defects on the road surface; finally, obtaining the defect opening path accurately characterizes the area with defects on the road surface, and road defect marking is performed based on the defect opening path, improving the accuracy of road defect identification. Attached Figure Description

[0012] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 The flowchart illustrates a machine vision-based method for identifying road surface defects, as provided in one embodiment of the present invention. Detailed Implementation

[0014] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a road surface defect identification method based on machine vision proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0015] Unless otherwise defined, 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 pertains.

[0016] The following description, in conjunction with the accompanying drawings, details a specific scheme for a machine vision-based road surface defect identification method provided by the present invention.

[0017] Please see Figure 1 The diagram illustrates a flowchart of a machine vision-based road surface defect identification method according to an embodiment of the present invention. The method includes the following steps: Step S1: Install the left and right lighting devices onto the inspection vehicle to obtain the left-side road surface image under the left-side lighting condition and the right-side road surface image under the right-side lighting condition.

[0018] In this embodiment of the invention, the implementation scenario is to identify road surface defects using machine vision technology to improve identification accuracy. First, an industrial camera is installed below the front of a road inspection vehicle, ensuring the camera's field of view covers the road surface area beneath the vehicle. Then, two lighting devices are installed on either side of the industrial camera. The left lighting device provides diagonal illumination from left to right, and the right lighting device provides diagonal illumination from right to left. The installation height, illumination distance, and illumination angle of the two lighting devices are symmetrical. The implementer can determine the installation method and adjust the illumination angle according to the implementation scenario. The left and right lighting devices use strobe or short-time synchronous illumination, with the illumination triggering time synchronized with the camera exposure time to reduce motion blur caused by vehicle movement. During vehicle operation, the controller collects data in the following sequence: The left lighting device is activated, the industrial camera captures an image of the road surface from the left, and the acquisition time and vehicle position are recorded; the right lighting device is activated, the industrial camera captures an image of the road surface from the right, and the acquisition time and vehicle position are recorded. After image acquisition, the two images are calibrated using camera calibration parameters to correct distortion and eliminate radial and tangential distortion of the lens.

[0019] Step S2: Perform position registration based on the grayscale difference features of the overlapping area of ​​the left and right road surface images; obtain the lighting response value of any position under different lighting conditions based on the local grayscale distribution features of any position in the left and right road surface images; obtain the lighting response difference value based on the difference features of the lighting response values ​​of any position under different lighting conditions and the standard lighting response difference features.

[0020] Because vehicles may undergo short displacements during the acquisition of road surface images with left and right lighting due to time differences, the same pixel positions in the left and right lighting road surface images do not need to correspond to the same road position. The positional correspondence between the two images needs to be adjusted through vehicle displacement recording and road plane transformation. To ensure sufficient overlap between the two images, the acquisition time interval must ensure that the vehicle displacement does not exceed half the effective overlap length of the two images. First, the plane containing the road surface is used as the road plane coordinate system. The horizontal axis of this coordinate system represents the left-right direction of the road, and the vertical axis represents the vehicle's travel direction. Based on the image capture area and the vehicle displacement during the acquisition interval between the two images, the coordinate positions of each pixel in the image are calibrated in the road plane coordinate system. The coordinate transformation process is achieved through a road plane transformation matrix, which is determined based on camera intrinsic parameters, camera mounting attitude, vehicle displacement, vehicle yaw angle, and pitch angle. When the vehicle experiences slight lateral shifts or attitude changes, the transformation matrix is ​​updated based on the vehicle yaw angle and pitch angle output by the inertial measurement unit, so that the images corresponding to both left and right lighting are transformed into the same road plane reference coordinate system. This coordinate transformation is existing technology, and the specific steps are not detailed here.

[0021] Furthermore, due to vehicle vibration, displacement sensor errors, and changes in vehicle attitude, the left and right road surface images after road plane transformation may still have slight offsets, causing the same coordinate in the coordinate system to represent different road surface positions in the two images. Therefore, fine registration is required to improve the accuracy of subsequent analysis. Thus, position registration is performed based on the grayscale difference characteristics of the overlapping area between the left and right road surface images. Preferably, in this embodiment of the invention, the position registration step includes:

[0022] In the road plane coordinate system, the right-side road surface image is fixed, while the left-side road surface image is moved; where, The lateral displacement of the left-facing road surface image is... Longitudinal displacement is The image's overall grayscale difference value at that time; This represents the set of reference locations in the overlapping area of ​​the left and right road surface images. Let N represent any reference position in the set of reference positions, and let N represent the number of positions in the set of reference positions. Indicates the reference position in the right-hand road surface image. grayscale value, The lateral displacement of the left-facing road surface image is... Longitudinal displacement is Time reference position The grayscale value. This represents the grayscale difference between the left illuminated road surface image and the original matching position of the right illuminated road surface image after the left illuminated road surface image has moved its coordinate position in the coordinate system. If the positions of the two images have been accurately matched, the grayscale difference at the same position in the left and right illuminated road surface images should be small. If the grayscale difference after the left illuminated road surface image is moved is even smaller, it indicates that there was a small offset in the initial coordinate registration. Furthermore, under different movement step sizes, the smaller the overall grayscale difference value of the images, the more likely the matching result after the movement can make the same coordinates in the two images represent the same road surface position. Therefore, the lateral displacement corresponding to the minimum value of the overall grayscale difference value of the images is... and longitudinal displacement are Move the left-facing road surface image and then register the right-facing road surface image with the moved left-facing road surface image; subsequent analysis will be performed based on the correspondence after the registration.

[0023] Furthermore, when road surface defects such as cracks and depressions occur, differences in brightness and darkness characteristics will occur when photographed under left and right lighting conditions. Because the asphalt color varies across different road sections, directly comparing the original grayscale can easily confuse the overall brightness and darkness variations of the road surface with local defects. Therefore, the illumination response value at any location under different lighting conditions is obtained based on the local grayscale distribution characteristics of any location in the left and right road surface images. Preferably, in this embodiment of the invention, the step of obtaining the illumination response value includes:

[0024] In the formula, k represents the lighting sign; when k is L, it represents the image of the road surface viewed from the left; when k is R, it represents the image of the road surface viewed from the right. This represents the lighting response value at any location i in the road surface image when the lighting identifier is k. This represents the grayscale value of any position i in the road surface image when the lighting indicator is k. This represents the median gray level of a preset neighborhood of any position i in the road surface image when the lighting identifier is k. This represents the absolute median deviation of the preset neighborhood of any position i in the road surface image when the lighting label is k. This indicates a preset minimum positive number. The local background grayscale at any given location is characterized by a preset neighborhood of 80 millimeters centered on that location, so that the neighborhood can include the damaged area and a portion of the surrounding normal road surface; the implementer can determine this neighborhood according to the implementation scenario. This indicates the difference between this arbitrary location and the local background. The median of the absolute value of the numerator is used for scaling the numerator. A very small positive number is preset to avoid a denominator of 0; in this embodiment, it is set to 0.001. The illumination response value represents the degree of brightness difference between the location and the local background; a smaller value means the location is darker relative to its surroundings, and vice versa.

[0025] After obtaining the illumination response values ​​of the same road surface location in the left-illuminated and right-illuminated road surface images, the difference in illumination response under the two illumination conditions is further analyzed. Therefore, the illumination response difference value is obtained based on the difference characteristics of the illumination response values ​​of this arbitrary location under different illumination conditions and the standard illumination response difference characteristics. Preferably, in this embodiment of the invention, the step of obtaining the illumination response difference value includes:

[0026] In the formula, This represents the difference in lighting response at any location i. This represents the illumination response value at any position i in the left-facing road surface image. This represents the illumination response value at any position i in the right-side road surface image. This represents the difference in illumination response values ​​at any position i between the left-facing and right-facing images of the road surface corresponding to a preset standard road surface. Since the optical power, installation angle, camera field of view, and road surface reflection conditions of the left and right lighting devices may differ, before formal testing, left-facing and right-facing images of a smooth, undamaged standard road surface are taken separately. The same processing steps are then performed to obtain the illumination response values ​​at the same position in both images. The difference in illumination response values ​​between the left-facing and right-facing images at the same position is then calculated to characterize the fixed differences caused by external factors. This represents the fixed external differences at any given location. Therefore, the illumination response difference value is the difference in illumination response between the left-illuminated road surface image and the right-illuminated road surface image after removing the fixed external differences. The smaller the value, the lower the left-illuminated response is relative to the right-illuminated response, and vice versa.

[0027] Step S3: Obtain the reverse response difference value based on the difference characteristics of the lighting response difference values ​​on both sides of any position at any angle; obtain the optimal reverse response difference value based on the difference characteristics of the reverse response difference values ​​at any position at different angles; obtain the local depression response point and the corresponding disease extension direction based on the data distribution characteristics of the optimal reverse response difference value at all positions; obtain the disease opening path based on the location distribution characteristics of the local depression response point and the disease extension direction.

[0028] When the left and right lighting switches, the light reception relationship on both sides of the depression or crack structure shifts, resulting in a significant difference in the lighting response values ​​on both sides of the defect. For example, if there is a crack in the direction of vehicle travel, left lighting will make the right side of the crack brighter and the left side darker, while right lighting will make the right side darker and the left side brighter. Therefore, the reverse response difference value can be obtained based on the difference characteristics of the lighting response values ​​on both sides at any position and any angle. Preferably, in this embodiment of the invention, the step of obtaining the reverse response difference value includes:

[0029] In the formula, Q represents the difference in the reverse response at any position in any angular direction, where the angular direction is defined as the angle with the horizontal axis of the coordinate system. When the value is 0 degrees, the angular direction is the horizontal axis direction. G represents the set of indices of the adjacent positions on both sides of the arbitrary position in this angular direction. This represents the difference in lighting response at the g-th position, starting from this arbitrary position and extending to any end of this arbitrary angular direction. This represents the difference in lighting response at the g-th position at the other end of the arbitrary angular direction. This represents a preset minimum positive number. Assuming the origin of the coordinate system is at this arbitrary position, and the horizontal axis is the direction of any angle, then... This represents the difference in lighting response at the g-th position along the positive or negative horizontal axis, starting from the origin. This represents the difference in lighting response at the g-th position in another direction, starting from the origin. If a crack exists in the vertical direction of the horizontal axis, the difference in lighting response at corresponding positions at both ends of the origin on the horizontal axis will be larger, thus leading to the derivation... Larger, and The difference in lighting response is relatively small; if there is no crack in the longitudinal direction, the difference in lighting response values ​​at corresponding positions on both sides of the transverse axis is relatively small; in this embodiment of the invention, the value of g is 15, that is, 15 adjacent positions on each side, which can be determined by the implementer according to the implementation scenario. Therefore, the larger the difference in reverse response value, the more likely there is a depression feature at that arbitrary position, and the direction of the depression extension is perpendicular to that arbitrary angle direction.

[0030] Furthermore, all angular directions at any given location can be traversed to obtain the reverse response difference values ​​corresponding to different angular directions, thereby obtaining the optimal reverse response difference value based on the difference characteristics of the reverse response difference values ​​at any given location in different angular directions. Preferably, in this embodiment of the invention, the step of obtaining the optimal reverse response difference value includes: taking the maximum value of the reverse response difference values ​​at any given location in different angular directions as the optimal reverse response difference value corresponding to that given location. When the optimal reverse response difference value characterizes the maximum probability of a depression feature at any given location, the vertical direction of the angular direction corresponding to the optimal reverse response difference value is more likely to be the extension direction of the depression opening. Furthermore, by traversing all locations in the overlapping area, local depression response points and corresponding disease extension directions can be obtained based on the data distribution characteristics of the optimal reverse response difference values ​​at all locations. Preferably, in this embodiment of the invention, the step of obtaining local depression response points and disease extension directions includes: taking the location where the optimal reverse response difference value exceeds a preset difference threshold as the local depression response point. In this embodiment of the invention, the preset difference threshold is the 75th percentile of the optimal reverse response difference values ​​at all locations in the overlapping area, which can be determined by the implementer according to the implementation scenario. The angle direction from which the best reverse response difference value of the local depression response point is obtained is taken as the target direction, and the direction perpendicular to the target direction is taken as the disease extension direction.

[0031] After obtaining the set of local depression response points in the overlapping area, the path of disease opening can be obtained based on the location distribution characteristics of the local depression response points and the direction of disease extension. Preferably, in this embodiment of the invention, the step of obtaining the path of disease opening includes: searching for other connected local depression response points along the direction of disease extension of the local depression response points with a preset search step size, traversing all local depression response points, and obtaining different path of disease opening. In this embodiment of the invention, the preset search step size is the average width of common cracks in historical detections, which can be determined by the implementer according to the implementation scenario. At positions that are at intervals of the search step size along the direction of disease extension of the local depression response point, other local depression response points within a preset search radius are searched; in this embodiment of the invention, the preset search radius is 10 mm, which can be determined by the implementer according to the implementation scenario. The direction of disease extension is two directions, and searches are performed at both ends; all local depression response points are traversed to obtain different path of disease opening. The opening path of the defect reflects common defects such as cracks, depressions, and potholes in the road surface; if there is an isolated local depression response point, it is not considered as a defect opening path, which may be due to interference from non-defect characteristics.

[0032] Step S4: Mark road defects according to the opening path of the defects.

[0033] After obtaining all the opening paths of road defects, road defects can be marked according to their locations, and inspection personnel can then carry out maintenance based on these marks. The opening paths can also be further optimized, for example, by filling closed opening paths to generate planar defect areas to characterize the size of potholes. Morphological operations are performed on crack-type opening paths to expand their width, giving the optimized defect areas area characteristics. Then, based on area and length characteristics, different levels of road defects are classified, and maintenance priorities are differentiated. Implementers can determine their own methods for marking road defects based on the opening paths; no specific limitations are imposed here. Finally, by analyzing the differences in the response of depressions, cracks, and potholes under different directional lighting conditions, and by analyzing the differences in lighting responses between left- and right-facing road images obtained by the inspection vehicle, the road defect paths are obtained, improving the accuracy of road defect detection.

[0034] In summary, this invention provides a machine vision-based method for identifying road surface defects. It performs position registration based on the grayscale difference features of left and right road surface images; obtains illumination response values ​​based on the local grayscale distribution features of any location in the left and right road surface images; obtains illumination response difference values ​​based on the difference features of illumination response values ​​under different illumination conditions and standard illumination response difference features; obtains the optimal reverse response difference value based on the difference features of illumination response values ​​on both sides of any location at any angle; and obtains local depression response points and defect extension directions based on the data distribution features of the optimal reverse response difference values ​​at all locations. This invention obtains defect opening paths based on the positional distribution features of local depression response points and defect extension directions; and marks road defects based on these opening paths, thus improving the accuracy of defect identification.

[0035] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0036] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for identifying road surface defects based on machine vision, characterized in that, The method includes the following steps: Install two lighting devices, one on the left and one on the right, onto the inspection vehicle to obtain images of the road surface under left-side lighting conditions and images of the road surface under right-side lighting conditions. Position registration is performed based on the grayscale difference characteristics of the overlapping area of ​​the left and right road surface images; the illumination response value of any location under different lighting conditions is obtained based on the local grayscale distribution characteristics of any location in the left and right road surface images; the illumination response difference value is obtained based on the difference characteristics of the illumination response values ​​of any location under different lighting conditions and the standard illumination response difference characteristics. The reverse response difference value is obtained based on the difference characteristics of the lighting response difference values ​​on both sides of any position at any angle; the optimal reverse response difference value is obtained based on the difference characteristics of the reverse response difference values ​​at different angles of the same position; the local depression response point and the corresponding disease extension direction are obtained based on the data distribution characteristics of the optimal reverse response difference values ​​at all positions; and the disease opening path is obtained based on the positional distribution characteristics of the local depression response point and the disease extension direction. Road defects are marked according to the opening paths of the defects.

2. The method for identifying road surface defects based on machine vision according to claim 1, characterized in that, The step of performing position registration based on the grayscale difference features of the overlapping area of ​​the left and right road surface images includes: In the road plane coordinate system, the right-side road surface image is fixed, and the left-side road surface image is moved. In the formula... The lateral displacement of the left-facing road surface image is... Longitudinal displacement is The image's overall grayscale difference value at that time; This represents the set of reference locations in the overlapping area of ​​the left and right road surface images. Let N represent any reference position in the set of reference positions, and let N represent the number of positions in the set of reference positions. Indicates the reference position in the right-hand road surface image. grayscale value, The lateral displacement of the left-facing road surface image is... Longitudinal displacement is Time reference position The grayscale value; the lateral displacement corresponding to the minimum value of the overall grayscale difference of the image is... and longitudinal displacement are Move the left-facing road surface image and then perform positional registration between the right-facing road surface image and the moved left-facing road surface image.

3. The method for identifying road surface defects based on machine vision according to claim 1, characterized in that, The step of obtaining the illumination response value of any location under different lighting conditions based on the local gray-level distribution features of any location in the left-illuminated road surface image and the right-illuminated road surface image includes: In the formula, k represents the lighting sign; when k is L, it represents the image of the road surface illuminated from the left; when k is R, it represents the image of the road surface illuminated from the right. This represents the lighting response value at any location i in the road surface image when the lighting identifier is k. This represents the grayscale value of any position i in the road surface image when the lighting indicator is k. This represents the median gray level of a preset neighborhood of any position i in the road surface image when the lighting identifier is k. This represents the absolute median deviation of the preset neighborhood of any position i in the road surface image when the lighting label is k. This indicates a preset minimum positive number.

4. The method for identifying road surface defects based on machine vision according to claim 1, characterized in that, The step of obtaining the lighting response difference value based on the difference characteristics of the lighting response values ​​at any location under different lighting conditions and the standard lighting response difference characteristics includes: In the formula This represents the difference in lighting response at any location i. This represents the illumination response value at any position i in the left-facing road surface image. This represents the illumination response value at any position i in the right-side road surface image. This represents the difference in illumination response values ​​at any position i between the left-facing and right-facing road surface images corresponding to the preset standard road surface.

5. The method for identifying road surface defects based on machine vision according to claim 1, characterized in that, The step of obtaining the reverse response difference value based on the difference characteristics of the illumination response difference values ​​on both sides at any position in any angular direction includes: In the formula, Q represents the difference in the reverse response at any position in any angular direction, and G represents the set of indices of adjacent positions on both sides of the arbitrary position in the arbitrary angular direction. This represents the difference in illumination response at the g-th position, taken as a starting point, at any end of the arbitrary angular direction. This represents the difference in illumination response at the g-th position at the other end of the arbitrary angular direction. This indicates a preset minimum positive number.

6. The method for identifying road surface defects based on machine vision according to claim 1, characterized in that, The step of obtaining the optimal reverse response difference value based on the difference characteristics of the reverse response difference values ​​at different angular directions at any position includes: The maximum value of the difference in the reverse response at any given position in different angular directions is taken as the optimal difference in the reverse response at that given position.

7. The method for identifying road surface defects based on machine vision according to claim 1, characterized in that, The step of obtaining the local depression response point and the corresponding disease extension direction based on the data distribution characteristics of the best reverse response difference value at all locations includes: The location where the optimal reverse response difference value exceeds a preset difference threshold is taken as the local depression response point; the angle direction in which the optimal reverse response difference value of the local depression response point is obtained is taken as the target direction, and the vertical direction of the target direction is taken as the disease extension direction.

8. The method for identifying road surface defects based on machine vision according to claim 1, characterized in that, The step of obtaining the opening path of the lesion based on the location distribution characteristics of the local depression response points and the direction of lesion extension includes: In the direction of disease extension of the local depression response point, other connected local depression response points are found with a preset search step size and connected. By traversing all local depression response points, different disease opening paths are obtained.

Images