A method for automated assessment of surface defects of a road marking

Abstract

A kind of automatic evaluation method of road marking surface defect, it includes the following steps: step 1, the image acquisition of the road marking of the detection section is carried out;Step 2, the collected marking image is preprocessed;Step 3, the main defect of road marking surface is identified, including crack, damage and pollution;Step 4, the severity of crack and damage is quantified;Step 5, the deterioration degree index S is calculated;Step 6, the surface condition is evaluated by comparing with marking surface condition evaluation rule table;Step 7, according to the deterioration degree index corresponding marking surface condition, marking maintenance suggestion is made, in addition, according to the deterioration degree index obtained at multiple times, the dynamic evolution curve of marking deterioration is formed.The automatic evaluation method of road marking surface defect provided by the application can avoid subjective error of artificial detection, and significantly improve detection speed and data processing efficiency;It is convenient for traffic management department to scientifically formulate marking maintenance scheme, and guarantees road safe operation.

Description

Technical Field

[0001] This invention relates to the field of highway maintenance technology, and also to the next generation of information technology fields with IPC classification numbers G06Q, G06V and G06T. Specifically, it involves using information technology to perform standardized automatic assessments of multiple indicators such as the degree of pollution, surface peeling area and cracking of road markings on highways, thereby providing a scientific basis for the engineering acceptance and maintenance decisions of road markings. Background Technology

[0002] Road markings are an important component of road traffic safety facilities, and their quality directly affects vehicle safety, traffic order, and road management. The functions of road markings include warning, instruction, and lane separation, playing a vital role in guiding traffic flow and ensuring driving safety. To ensure the performance and lifespan of road markings, regular quality inspections and maintenance assessments are necessary.

[0003] Currently, the quality inspection of road markings in my country mainly relies on manual visual inspection. Based on the existing GB / T 16311-2009 "Quality Requirements and Inspection Methods for Road Traffic Markings" and related industry standards, on-site inspectors typically use inspection forms to evaluate items such as color, clarity, whether the markings are peeling off, whether they are contaminated, and whether the edges are straight. Many inspection items are only categorized into two results: "present / absent" and "qualified / unqualified," lacking scientific, quantitative, and systematic evaluation standards.

[0004] The existing manual inspection methods have the following main limitations:

[0005] Highly subjective: Different testing personnel are influenced by factors such as experience and aesthetics, resulting in significant subjective differences in test results, making it difficult to achieve unified standards and objective comparability.

[0006] Low efficiency: Manual inspection is inefficient, especially when batch inspection or rapid acceptance is required, it is difficult to meet actual needs and is prone to omissions and oversights.

[0007] Lack of systematic data management: During manual testing, the test data is mostly recorded on paper, lacking centralized and standardized digital management, which makes it difficult to store and analyze the data in a unified manner, affecting subsequent quality evaluation and scientific decision-making.

[0008] Safety hazards: Conducting manual inspections in high-traffic environments poses certain risks to personal safety.

[0009] With the development of information technologies such as digital imaging, image processing, and computer vision, the use of automated and intelligent methods for road marking quality inspection has become an industry trend. Some sectors, such as manufacturing and electronic inspection, have widely applied computer vision to achieve efficient and objective automated inspection. In the field of traffic engineering, automated and quantitative inspection technologies for road marking quality, including appearance characteristics such as pollution levels, surface peeling area, and cracking, have also seen some development. For example, Chinese patent CN115619972A provides a "Road Marking Wear Identification and Evaluation Method Based on Point Cloud Data." Its technical solution is based on designing and optimizing a deep learning network model based on road marking features to extract road marking point clouds; based on differences in road marking shape, it constructs multiple shape descriptors to distinguish different types of road markings; it automatically calculates the wear area of ​​road markings, establishes wear assessment indicators, and grades and evaluates the degree of road marking wear. This technical solution uses point cloud data collected by LiDAR technology to extract and calculate road marking data. If road construction or management parties use this technical solution, they usually need to purchase data collected by map data vendors, which can lead to excessive costs.

[0010] Furthermore, Chinese patent CN117746366A provides a technical solution for "Road Marking Evaluation Method, Device, Equipment, Storage Medium, and Program Product." The evaluation method involves: acquiring the application scenario of the road marking and the error values ​​of multiple target points on each road marking; determining at least one level of accuracy threshold based on the application scenario; for any given road marking, calculating multiple sets of membership results based on the error values ​​of the target points and the accuracy threshold at least one level using a pre-constructed membership function; each set of membership results includes at least one membership degree; for each set of membership results, determining whether the membership result meets preset conditions based on the value of at least one membership degree; if so, evaluating the accuracy value of the road marking based on the membership results and preset membership principles. This approach integrates the error values ​​of all points on the road marking, considers the membership degrees of all error values, and comprehensively evaluates the accuracy value of the road marking, thereby improving the accuracy of road marking geometric accuracy assessment. This technical solution requires data sampling on the road markings of the target road to obtain real data for multiple target points, which is then compared with point cloud data to obtain the target point error value. Therefore, it does not have engineering application value for scenarios such as highways. In other words, for roads outside urban areas (including county roads, township roads, etc.), especially highways, the existing automated road marking evaluation methods are not yet fully mature, and the acceptance and maintenance of road markings in related road engineering projects still mainly rely on manual experience judgment. Summary of the Invention

[0011] In view of the shortcomings of the existing technology, the purpose of this invention is to provide an automated assessment method for surface defects of road markings, which can realize a rapid, objective and accurate assessment of the surface condition of road markings through image data, so as to improve the automation level of road marking acceptance and maintenance work, thereby improving the scientific nature and efficiency of road marking maintenance work.

[0012] To achieve the above objectives, the present invention provides an automated assessment method for surface defects in road markings, comprising the following steps:

[0013] Step 1: Use a camera with a resolution of at least 1920×1080 pixels to collect images of the road markings on the detection section.

[0014] Step 2 involves preprocessing the acquired line marking images, including noise reduction, contrast enhancement, environmental impact elimination, geometric correction, and format standardization. This includes using filtering algorithms to suppress noise, employing local histogram equalization to enhance image contrast, and utilizing illumination compensation algorithms to ensure uniformity of image brightness and color consistency.

[0015] Step 3: For the image preprocessed in Step 2, image processing technology is used to automatically identify the main defects on the road marking surface, including cracks, damage, and pollution, to achieve accurate classification and location of defects. This includes applying edge detection and morphological methods to detect through cracks and damage, judging pollution based on low saturation, low brightness, and grayscale thresholds set in the HSV color space, excluding damaged areas, completing the segmentation of polluted areas, obtaining the pollution area ratio that reflects the size of the pollution range, and normalizing the saturation pollution index, brightness pollution index, and grayscale pollution index. The saturation pollution index characterizes the deviation of the average saturation of the polluted area from the saturation of the ideal clean area. The brightness pollution index is used to measure the decrease in the average brightness of the polluted area relative to the normal brightness. The grayscale pollution index reflects the relative position of the grayscale level of the polluted area relative to the clean area and the severely polluted area. The pollution degree index is calculated by multiplying the pollution area ratio by the maximum value of the above three pollution indices.

[0016] Step 4: Based on the identification results of Step 3, quantify the severity of cracks and damage based on the number of cracks and the proportion of damage, that is, calculate the normalized index of the number of cracks and the normalized index of the proportion of damage area.

[0017] Step 5: Based on the calculation results of Step 3 and Step 4, the normalized index of crack quantity, the normalized index of damaged area ratio, and the pollution degree index are weighted and summed according to preset weights to calculate the deterioration degree index S, thereby quantifying the deterioration degree of the road marking surface.

[0018] Step 6: Based on the quantitative analysis results from Step 5, i.e., the deterioration index S, evaluate the surface condition according to the surface condition evaluation rules for the road markings. The surface condition evaluation rules for the road markings are as follows:

[0019] If the degradation index S < 0.1, it indicates that the surface condition of the marking is good.

[0020] If the degradation index is 0.1 ≤ S < 0.3, it indicates that the surface condition of the marking is average.

[0021] If the degradation index is 0.3 ≤ S < 0.5, it indicates that the surface condition of the markings is poor.

[0022] If the degradation index S≥0.5, it indicates that the surface condition of the marking is severely degraded.

[0023] Step 7: Store the test data and deterioration index results, and formulate road marking maintenance recommendations based on the surface condition of the road markings corresponding to the deterioration index. In addition, based on the deterioration index obtained at multiple time points, form a dynamic evolution curve of road marking deterioration to provide a basis for road marking maintenance decisions.

[0024] Preferably, in step 1, the camera device is an industrial camera with a resolution of not less than 1920×1080 pixels, equipped with a CMOS image sensor with a pixel count of not less than 2 million; the camera device is fixedly installed on the roof of the inspection vehicle at a height of 1-1.5 meters, and the camera angle is adjusted to 30°-45° according to the width of the road being inspected. When acquiring images, the vehicle speed is controlled at 60 to 80 km / h. For the selected acquisition line, a static image is obtained by extracting frames from the dynamic video, and each image contains a clear line length of 4-6 meters.

[0025] Preferably, in step 2, a two-dimensional Gaussian filtering algorithm is first used to smooth the image, effectively suppressing noise while preserving edge details as much as possible. Then, a local histogram equalization method is used to divide the image into multiple local regions, independently calculating the gray-level histogram of each region and performing equalization processing to enhance image contrast and highlight the gray-level difference between the markings and the background. Next, an improved Retinex illumination compensation algorithm is used to decompose the image into reflectance and illumination components, adjusting the illumination component to eliminate the influence of ambient light changes and improve image brightness uniformity and color stability. Subsequently, geometric and perspective corrections are performed on the image to eliminate distortion caused by the camera angle, ensuring the accuracy of the marking shape. Finally, the image resolution and format are uniformly adjusted to complete format standardization.

[0026] Preferably, in step 3, for crack detection, firstly, an edge detection operator is applied to extract the crack edge, capturing the parts of the image with obvious grayscale changes; then, a morphological thinning algorithm is used to thin the crack edge to a single pixel width to improve positioning accuracy; area filtering is used to remove noisy small regions; then, through connected component analysis, through-cracks are selected based on the ratio of the vertical length of the connected region to the image height to ensure that only valid cracks are retained; finally, morphological dilation and erosion operations are used to connect the crack breakpoints to form a continuous and complete crack mask.

[0027] Preferably, in step 3, after smoothing the grayscale image using a two-dimensional Gaussian filter to reduce noise interference, edge detection is performed using the Canny operator with a threshold set to 0.2 to 0.3 to capture crack edges with significant grayscale changes. Subsequently, a morphological thinning algorithm is used to refine the edges to a single pixel width, while area filtering is applied to remove small noise regions. Combined with connected component analysis, through cracks with a vertical length accounting for at least 0.8% of the image height are selected to effectively preserve the true crack portion. Then, morphological dilation and erosion operations are used to connect the crack breakpoints, forming a continuous and complete crack mask, and the number of cracks is counted. .

[0028] Preferably, in step 3, for damage detection, edge detection is first performed to identify the boundary of the damaged area; morphological dilation and closure operations are applied to bridge the gaps between the edges and realize the connectivity of the damaged area; by calculating the area of ​​the connected region, small areas smaller than a predetermined threshold are eliminated, and larger and more significant damaged areas are retained; finally, a damage mask that accurately represents the location of the damage is generated.

[0029] Preferably, in step 3, Canny edge detection is applied to the grayscale image with a threshold range of 0.1 to 0.3. Then, dilation and closure operations are performed to close edge cracks, with structuring element radii of 2 and 3, respectively. Through connected region area analysis, small regions smaller than 1500 pixels are removed, retaining only significantly damaged areas to generate a mask accurately reflecting the damage location. This mask is then surrounded by a curve in the tracing image. The proportion of damaged area is calculated based on the ratio of the number of pixels in the damaged mask to the total number of pixels in the image. In the formula Indicates the percentage of damaged area. This indicates the number of pixels marked as damaged in the damage mask. This represents the total number of pixels in the image.

[0030] Preferably, in step 3, for contamination detection, the preprocessed image is first converted to the HSV color space, and the saturation and brightness channels are extracted; based on preset threshold conditions, pixel areas with low saturation, low brightness, and low grayscale are screened to initially determine contamination candidate areas; then, the contamination candidate areas are logically NOTed with the damaged mask to eliminate the interference of damaged areas on contamination detection; finally, a clean contamination mask is obtained for subsequent quantitative analysis of the degree of contamination.

[0031] Preferably, in step 3, the color image is first converted to the HSV color space, and the saturation and brightness channels are extracted. The saturation threshold is set to 0.08, the brightness threshold to 0.97, and the grayscale threshold to 245. Pixel areas that meet the criteria of saturation below 0.08, brightness below 0.97, and grayscale below 245 are initially identified as candidate contamination areas. To avoid interference from damaged areas, a logical NOT operation is used to remove areas covered by the damaged mask, obtaining a clean contamination mask. Then, the average grayscale, average saturation, and average brightness of the contamination areas are calculated, and the saturation contamination index is calculated for each. Brightness pollution index and gray pollution index .

[0032] Saturation pollution index The deviation between the average saturation of the polluted area and the saturation of the ideal clean area is characterized by the following calculation method: ,in, It is the average saturation within the contamination mask area; the lower the saturation, the lower the saturation. The closer the value is to 1, the more severe the pollution; the higher the saturation, the closer this index is to 0, indicating slight or no pollution.

[0033] Brightness Pollution Index The method used to measure the decrease in average brightness of a polluted area relative to normal brightness is as follows: ,in, It is the average brightness of the polluted area; when the brightness is low, A higher value indicates that pollution is causing the surface to become dull; a higher brightness value indicates... A lower value indicates a smaller impact from pollution.

[0034] Gray pollution index The gray level of the contaminated area, relative to clean and heavily contaminated areas, is calculated as follows: ,in, C is the average gray value of the contamination mask area. gray This represents the maximum grayscale value of the clean marking area, indicating the upper limit of grayscale value in the normal, uncontaminated marking area; P grayThis represents the minimum gray level of the polluted area, indicating the lower limit of gray level in the severely polluted region. The gray pollution index is normalized to a range of 0 to 1. The lower the value, the more severe the pollution.

[0035] Next, the pollution level index L, which reflects the degree of pollution's impact on the marking surface, is calculated. pollution The calculation method is as follows: ,in, The percentage of polluted area, which represents the extent of pollution, is calculated as follows: , This indicates the number of pixels marked as contaminated in the contamination mask. This represents the total number of pixels in the image.

[0036] Preferably, in step 4, the number of cracks and the proportion of damaged area are normalized, converting these two indicators to a uniform range of 0 to 1. First, the normalized index for the number of cracks is calculated. , ,in, This is the maximum number of through cracks allowed; exceeding this value is considered severe cracking of the road markings. The maximum number of through cracks is determined based on the road grade. For highways and national / provincial roads, where the integrity and safety requirements for road markings are high, three or more through cracks are considered severe cracking. For lower-grade roads, five or more through cracks are considered severe cracking.

[0037] Then calculate the normalized index of the damaged area ratio. , ,in This is the damage rate threshold; exceeding this value indicates severe damage to the road markings. The damage rate threshold is set based on the functional requirements, safety needs, and maintenance experience of the road markings.

[0038] Preferably, It is 0.03.

[0039] Preferably, in step 5, the normalized index of the number of cracks obtained in step 4 is used. Normalized index of damaged area ratio Pollution level index L pollution The deterioration index S of the road marking surface is calculated by weighted summation according to preset weights, as shown in the following formula:

[0040]

[0041] Among them, the preset weights are respectively used , , This indicates that the sum of the weights is 1.

[0042] Preferably, in step 5, , , .

[0043] Preferably, in step 7, the dynamic evolution curve of the road marking deterioration trend and the proposal of maintenance suggestions are achieved through the following steps:

[0044] S701 collects image data of road markings on fixed dates each month;

[0045] S702, following steps 2 to 6, the surface deterioration index S of the marking is obtained;

[0046] S703, the degradation index S obtained each month is combined into a time series to form a dynamic evolution curve of the degradation of the line;

[0047] S704, using time series analysis to eliminate the impact of short-term fluctuations, determines the upward, stable, or downward trend of the deterioration index S;

[0048] Based on the deterioration index S and its changing trend, maintenance recommendations are proposed for S705:

[0049] The degradation index is low and shows a stable or declining trend. The maintenance recommendation is to conduct regular inspections and data monitoring to maintain the current condition.

[0050] The degradation index is moderate but continues to rise. The maintenance recommendation is timely maintenance, including damage repair and surface coating.

[0051] The degradation index is high and shows a stable or upward trend, so emergency maintenance is recommended.

[0052] Preferably, in step 7, the time series analysis method is a moving average, trend line fitting, or regression analysis.

[0053] The automated evaluation method for surface defects of road markings provided by this invention realizes automated and refined evaluation of road marking quality, avoids subjective errors in manual inspection, and can significantly improve inspection speed and data processing efficiency; it also facilitates traffic management departments to scientifically formulate road marking maintenance plans and ensure safe road operation. Attached Figure Description

[0054] Figure 1 This is an image diagram illustrating the road marking damage identification result of an automated assessment method for surface defects of road markings according to a specific embodiment of the present invention.

[0055] Figure 2 According to Figure 1 An image illustration of the road marking pollution identification results from an automated assessment method for surface defects in road markings. Detailed Implementation

[0056] To provide a clearer understanding of the technical features, objectives, and effects of the present invention, specific embodiments of the present invention will now be described in conjunction with the accompanying drawings.

[0057] This invention provides an automated method for evaluating surface defects in road markings, comprising the following steps:

[0058] Step 1: Use high-resolution camera equipment to capture images of the road markings on the inspection section to ensure rich image details and accurately reflect the surface condition of the markings.

[0059] Specifically, the camera device can be an industrial camera with a resolution of no less than 1920×1080 pixels, equipped with a CMOS image sensor and a pixel count of no less than 2 million. The industrial camera is fixedly mounted on the roof of the inspection vehicle at a height of 1-1.5 meters to avoid the vehicle itself affecting the image area during shooting. The camera's viewing angle can be adjusted to approximately 30°-45° depending on the width of the road being inspected to cover the entire road marking area. During image acquisition, the vehicle speed can be controlled at approximately 60-80 km / h to ensure clear and stable images even at high speeds. To adapt to different lighting environments (such as tunnels or rainy days), a high-brightness LED auxiliary light source can be equipped to compensate for low light levels at night and in shadow areas. The auxiliary light source can be a commercially available product that supports automatic control, including automatic on / off and brightness adjustment functions, allowing for automatic adjustment of auxiliary lighting based on ambient light intensity. The camera device can be equipped with autofocus and image stabilization to ensure distortion-free or blur-free image acquisition.

[0060] In a specific embodiment of this invention, a 10-kilometer section of a highway in Guizhou Province was selected for testing. The main reason for choosing this target section is that the climate along this section is variable and traffic is frequent, making road markings susceptible to wear and tear from the natural environment and traffic, leading to accelerated surface deterioration and noticeable cracks, damage, and contamination. Using the evaluation method of this invention, high-resolution images of the road markings can be collected to comprehensively and objectively assess the surface condition, accurately identify various defects, and quantitatively analyze their severity, providing a scientific basis for subsequent maintenance decisions. This section is a three-lane road in one direction; the markings between the slow lane and the middle lane were selected for collection because the center line is subject to higher vehicle traffic frequency, resulting in more pronounced surface deterioration. For the selected markings, static images were obtained by extracting frames from dynamic video. Each image contains a clear section of marking 4-6 meters in length, ensuring that over 90% of the image data along the overall length of the road markings is obtained during the entire data collection process.

[0061] In this embodiment, an industrial camera with a resolution of 1920×1080 and 2 megapixels is selected and fixedly mounted on the roof of the inspection vehicle at a height of 1.2 meters. The camera's viewing angle is adjusted to 40°, and the vehicle's operating speed is maintained at approximately 80 km / h. The industrial camera continuously captures dynamic images of the marked area in video format. Through subsequent image frame extraction and filtering, high-quality still photos are extracted from the video for analysis. The video frame extraction is dynamically adjusted according to the vehicle's speed; specifically, the extraction frequency is no less than four frames per second, thus ensuring sufficient coverage of the 6-meter-long area along the marked line. After image frame extraction, blurry, abnormally exposed, occluded, and optically distorted image frames are discarded. Manual verification is then performed to ensure that the selected images are clear, accurately focused, and free of significant reflections in the marked area. The final high-quality still photos are used for subsequent precise defect analysis and evaluation.

[0062] Step 2 involves preprocessing the acquired line marking images, including noise reduction, contrast enhancement, environmental impact elimination, geometric correction, and format standardization. This includes using filtering algorithms to suppress noise, employing local histogram equalization to enhance image contrast, and utilizing illumination compensation algorithms to ensure uniformity of image brightness and color consistency.

[0063] Specifically, the image is first smoothed using a two-dimensional Gaussian filtering algorithm, effectively suppressing noise while preserving edge details as much as possible. Then, a local histogram equalization method is used to divide the image into multiple local regions, independently calculating and equalizing the gray-level histogram of each region to enhance image contrast and highlight the gray-level difference between the markings and the background. Next, an improved Retinex illumination compensation algorithm is used to decompose the image into reflectance and illumination components, adjusting the illumination component to eliminate the influence of ambient light variations and improve image brightness uniformity and color stability. Subsequently, geometric and perspective corrections are performed to eliminate distortions caused by camera angles, ensuring accurate marking shape. Finally, the image resolution and format are standardized to provide a standardized input for subsequent automatic detection of cracks, damage, and contamination.

[0064] Step 3: For the image preprocessed in Step 2, image processing technology is used to automatically identify the main defects on the road marking surface, including cracks, damage, and pollution, to achieve accurate classification and location of defects. This includes using edge detection and morphological methods to detect through cracks and damage, judging pollution based on low saturation, low brightness, and grayscale thresholds set in the HSV color space, excluding damaged areas, completing the segmentation of polluted areas, and calculating the pollution degree index.

[0065] Specifically, for the road marking images preprocessed in step 2, image processing technology is used to automatically identify major defects, including cracks, damage, and pollution, to achieve accurate classification and location.

[0066] In crack detection, an edge detection operator is first applied to extract crack edges, capturing areas with significant grayscale changes in the image. A morphological thinning algorithm is then used to refine the crack edges to a single pixel width to improve positioning accuracy. An area filtering method is used to remove noisy small regions. Next, through connected component analysis, through-cracks are selected based on the ratio of the vertical length of the connected region to the image height, ensuring that only valid cracks are retained. Finally, morphological dilation and erosion operations are used to connect crack breakpoints, forming a continuous and complete crack mask.

[0067] Specifically, after smoothing the grayscale image using a two-dimensional Gaussian filter to reduce noise interference, the Canny operator is used for edge detection with a threshold set between 0.2 and 0.3 to capture crack edges with significant grayscale changes. Then, a morphological thinning algorithm is used to refine the edges to a single pixel width, while area filtering is applied to remove small noise regions. Combined with connected component analysis, through cracks with a vertical length accounting for at least 0.8 of the image height are selected to effectively preserve the true crack portions. Finally, morphological dilation and erosion operations are used to connect the crack breakpoints, forming a continuous and complete crack mask, and the number of cracks is counted. In this embodiment, no sufficiently long and continuous cracks were detected, therefore the statistical result for through cracks is 0, i.e. .

[0068] Figure 1 This is an image diagram illustrating the road marking damage identification result of an automated assessment method for road marking surface defects according to a specific embodiment of the present invention; see also Figure 1 As shown, in damage detection, edge detection is first performed to identify the boundaries of the damaged area; morphological dilation and closure operations are applied to bridge the gaps between the edges and achieve connectivity of the damaged area; by calculating the area of ​​the connected region, small areas smaller than a predetermined threshold are eliminated, and larger and more significant damaged areas are retained; finally, a damage mask that accurately represents the location of the damage is generated.

[0069] Specifically, Canny edge detection is applied to the grayscale image with a threshold range of 0.1 to 0.3. Dilation and closing operations are then performed to close edge cracks, with structuring element radii of 2 and 3, respectively. Through connected component area analysis, small regions smaller than 1500 pixels are removed, retaining only significantly damaged areas. A mask accurately reflecting the damage location is generated, and this mask is enclosed by a curve in the tracing image. The proportion of damaged area is calculated based on the ratio of the number of pixels in the damaged mask to the total number of pixels in the image. In the formula This indicates the number of pixels marked as damaged in the damage mask. This represents the total number of pixels in the image. In this implementation, .

[0070] Figure 2 According to Figure 1 An image illustration of the road marking contamination identification results from an automated assessment method for road marking surface defects, see [link / reference]. Figure 2 As shown, in contamination detection, the preprocessed image is first converted to the HSV color space, and the saturation and brightness channels are extracted. Based on preset threshold conditions, pixel areas with low saturation, low brightness, and low grayscale are screened to initially determine contamination candidate areas. Subsequently, a logical NOT operation is performed between the contamination candidate areas and the damaged mask to eliminate the interference of damaged areas on contamination detection. Finally, a clean contamination mask is obtained for subsequent quantitative analysis of the degree of contamination.

[0071] Specifically, the color image is first converted to the HSV color space, and the saturation and brightness channels are extracted. The saturation threshold is set to 0.08, the brightness threshold to 0.97, and the grayscale threshold to 245. Pixel areas with saturation below 0.08, brightness below 0.97, and grayscale below 245 are initially identified as candidate contamination areas. To avoid interference from damaged areas, a logical NOT operation is used to remove areas covered by the damaged mask, obtaining a clean contamination mask. Next, the average grayscale, average saturation, and average brightness of the contamination areas are calculated, and the saturation contamination index is calculated for each. Brightness pollution index and gray pollution index .

[0072] Saturation pollution index The deviation between the average saturation of the polluted area and the saturation of the ideal clean area is characterized by the following calculation method: ,in, This is the average saturation within the contamination mask area. The lower the saturation, the lower the saturation. The closer the value is to 1, the more severe the pollution; the higher the saturation, the closer this index is to 0, indicating slight or no pollution.

[0073] Brightness Pollution Index The method used to measure the decrease in average brightness of a polluted area relative to normal brightness is as follows: ,in, This is the average brightness of the polluted area. When the brightness is low, A higher value indicates that pollution is causing the surface to become dull; a higher brightness value indicates... A lower value indicates a smaller impact from pollution.

[0074] Gray pollution index The gray level of the contaminated area, relative to clean and heavily contaminated areas, is calculated as follows: ,in, C is the average gray value of the contamination mask area. gray This represents the maximum grayscale value of the clean marking area, indicating the upper limit of grayscale value in the normal, uncontaminated marking area; P gray This represents the minimum gray level of the polluted area, indicating the lower limit of gray level in the severely polluted region. The gray pollution index is normalized to a range of 0 to 1. The lower the value, the more severe the pollution.

[0075] Then, the contamination level index L, which reflects the degree of contamination on the surface of the markings, can be calculated. pollution The calculation method is as follows: ,in, The percentage of polluted area, which represents the extent of pollution, is calculated as follows: , This indicates the number of pixels marked as contaminated in the contamination mask. This represents the total number of pixels in the image.

[0076] In this implementation, the grayscale threshold C of the clean area is taken respectively. gray The value is 250, and the pollution grayscale threshold P is... gray The pixel ratio R of the polluted area in the datum image is 50. pollution The average saturation is 0.1566. The average brightness is 0.0728. The average gray level is 0.9458. The value is 238.19, therefore,

[0077] ,

[0078] ,

[0079] .

[0080] Pollution level index L pollution The calculation method is the product of the proportion of pixels in the contaminated area and the maximum value of the three contamination indices mentioned above:

[0081] .

[0082] Step 4: Based on the identification results of Step 3, quantify the severity of cracks and damage based on the number of cracks and the proportion of damage, that is, calculate the normalized index of the number of cracks and the normalized index of the proportion of damage area.

[0083] Specifically, the number of cracks and the proportion of damaged area are normalized to a uniform range of 0 to 1, avoiding unreasonable weighting in the comprehensive evaluation due to differences in numerical dimensions and ranges, thereby ensuring the scientific and fair nature of subsequent weighted calculations.

[0084] First, calculate the normalized index of crack number. , ,in, The maximum number of through cracks is set; exceeding this value is considered severe cracking of the road markings. The principle for setting the maximum number of through cracks is based on the road grade. For highways and national / provincial trunk roads, due to their higher requirements for the integrity and safety of road markings, three or more through cracks are considered severe cracking. For lower-grade roads, five or more through cracks are considered severe cracking. In this embodiment... It is 3. ,therefore =0.

[0085] Then calculate the normalized index of the damaged area ratio. , ,in A damage rate threshold is set; exceeding this value indicates severe damage to the road markings. This threshold is established based on the functional requirements, safety needs, and maintenance experience of the road markings. Specifically, for the road section selected in this embodiment, maintenance experience shows that when the damaged area exceeds 3% of the total marking area, the visual continuity and recognition effectiveness of the markings significantly decrease, potentially leading to driver misjudgment or ignoring of marking information, increasing traffic safety risks. Therefore, in this embodiment… It is 0.03. ,therefore =1.

[0086] Step 5: Based on the calculation results of Step 3 and Step 4, calculate the deterioration index S to quantify the degree of deterioration of the road marking surface.

[0087] Specifically, the normalized index of the number of cracks obtained in step 4 Normalized index of damaged area ratio Pollution level index L pollution The deterioration index S of the road marking surface is calculated by weighted summation according to preset weights, as shown in the following formula:

[0088]

[0089] Among them, the preset weights are respectively used , , This indicates that the sum of the weights is 1. Since the number of cracks, the proportion of damaged area, and the degree of contamination are normalized to the same scale, a total weight of 1 ensures that each indicator contributes proportionally and reasonably to the comprehensive evaluation, avoiding any one indicator from excessively dominating the evaluation result, and ensuring that the evaluation result comprehensively reflects the condition of the marking line. In this embodiment, , , Damage usually directly leads to the destruction of the road marking structure, with a significant impact, and therefore has a higher weight; cracks, although defects, have a relatively smaller impact on visual recognition; contamination affects clarity but does not damage the structure, so both have relatively lower weight.

[0090] therefore,

[0091]

[0092] Step 6: Based on the quantitative analysis results of Step 5, namely the deterioration index, evaluate the surface condition by referring to the surface condition evaluation rule table of the marking line.

[0093] Specifically, the degradation index S ranges from 0 to 1. A value closer to 0 indicates fewer cracks, a lower damage rate, and less contamination on the road marking surface, indicating a good overall surface condition. Conversely, a value closer to 1 indicates more cracks, larger damaged areas, and more severe contamination on the road marking surface, indicating severe overall surface degradation. Based on the road marking surface condition evaluation table, it can be concluded that in this embodiment, S = 0.5363, which is greater than 0.5, indicating severe degradation of the road marking surface.

[0094] The evaluation rules for the surface condition of road markings are as follows:

[0095] .

[0096] Step 7: Store the test data and deterioration index results, and formulate road marking maintenance recommendations based on the surface condition of the road markings corresponding to the deterioration index. In addition, based on the deterioration index obtained at multiple time points, form a dynamic evolution curve of road marking deterioration to provide a basis for road marking maintenance decisions.

[0097] Specifically, by establishing a data management platform, all detection data, defect classifications, scoring indicators, and collected spatiotemporal information can be automatically saved in real time. This allows for dynamic assessment of the deterioration trend of the road markings through comparative analysis of data from multiple time periods, providing a basis for maintenance cycle planning.

[0098] The dynamic evolution curve of road marking deterioration trend and the proposal of maintenance recommendations can be achieved through the following steps:

[0099] The S701 uses an industrial camera to capture high-resolution images of road markings on a fixed date each month (e.g., the 1st of each month).

[0100] S702, the aforementioned steps 2 to 6, yield the surface deterioration index S of the marking line;

[0101] S703, the degradation index S obtained each month is combined into a time series to form a dynamic evolution curve of the degradation of the line;

[0102] S704 uses time series analysis methods, such as moving average, trend line fitting, and regression analysis, to eliminate the influence of short-term fluctuations and determine the upward, stable, or downward trend of the deterioration index S.

[0103] Based on the deterioration index S and its changing trend, maintenance recommendations are proposed for S705:

[0104] The degradation index is low and shows a stable or declining trend. Regular inspections and data monitoring are recommended to maintain the status quo.

[0105] The degradation index is moderate but continuously rising, so timely maintenance is recommended, including damage repair and surface coating.

[0106] The degradation index is high and shows a stable or upward trend, so emergency maintenance is recommended.

[0107] In the specific embodiment involved in this application, the comprehensive deterioration index S=0.5363 calculated in step 6 exceeds the threshold of 0.5. Therefore, the surface condition of the road markings can be assessed as "severely deteriorated," which prompts the maintenance party to take emergency maintenance measures and repaint the markings to maintain their function. This information, along with the number of cracks, the proportion of damaged area, the pollution index, and the corresponding image analysis results, can be stored together to facilitate subsequent analysis of the changing trends of road markings on this road section.

[0108] After road markings are applied to newly constructed roads or during the maintenance of existing roads, images of the markings can be collected periodically at fixed intervals (e.g., on the 1st of each month) to conduct surface condition assessments in order to continuously monitor their condition. Based on time series analysis and dynamic evolution curves, the deterioration trend of the markings can be determined, and maintenance recommendations can be made in a timely manner to ensure road driving safety and the long-term effectiveness of the markings.

[0109] The automated evaluation method for surface defects of road markings provided by this invention realizes automated and refined evaluation of road marking quality, avoids subjective errors in manual inspection, and can significantly improve inspection speed and data processing efficiency; it also facilitates traffic management departments to scientifically formulate road marking maintenance plans and ensure safe road operation.

[0110] Those skilled in the art should understand that although the present invention has been described with reference to multiple embodiments, not every embodiment contains only one independent technical solution. This description is provided merely for clarity; those skilled in the art should understand the specification as a whole and consider the technical solutions involved in each embodiment as being able to be combined with each other to form different embodiments to understand the scope of protection of the present invention.

[0111] The above description is merely an illustrative embodiment of the present invention and is not intended to limit the scope of the invention. Any equivalent changes, modifications, and combinations made by those skilled in the art without departing from the concept and principles of the present invention should fall within the scope of protection of the present invention.

Claims

1. A method for automated assessment of surface defects of a road marking, characterized in that It comprises the following steps: Step 1, using a camera device with resolution not less than 1920x1080 pixels to collect images of the road markings of the detection section; Step 2, pre-processing the collected marking images, including noise removal, contrast enhancement, environmental influence elimination, geometric correction and format standardization, wherein a filtering algorithm is used to suppress image noise, a local histogram equalization method is used to enhance image contrast, and an illumination compensation algorithm is used to ensure image brightness uniformity and color consistency; Step 3, using image processing technology to automatically identify the main defects of the road marking surface, including cracks, damage and pollution, to achieve accurate classification and positioning of defects, including applying edge detection and morphological methods to detect through cracks and damage, determining pollution based on the low saturation, low brightness and grayscale threshold set in the HSV color space, excluding damaged areas, segmenting the pollution area, obtaining the pollution area ratio reflecting the size of the pollution area, the normalized saturation pollution index, brightness pollution index and grayscale pollution index, the saturation pollution index representing the deviation of the average saturation of the pollution area from the saturation of the ideal clean area, the brightness pollution index for measuring the decline of the average brightness of the pollution area relative to the normal brightness, and the grayscale pollution index reflecting the relative position of the grayscale level of the pollution area relative to the clean area and the severely polluted area, and calculating the pollution degree index according to the product of the pollution area ratio and the maximum value of the above three pollution indexes; Step 4, according to the identification results of step 3, quantifying the severity of cracks and damage based on the number of cracks and the proportion of damage, i.e. calculating the crack number normalization index and the damage area proportion normalization index; Step 5, according to the calculation results of step 3 and step 4, weighting and summing the crack number normalization index, damage area proportion normalization index and pollution degree index according to the pre-set weight to calculate the deterioration degree index S, and quantifying the deterioration degree of the road marking surface; Step 6, according to the quantification analysis result of step 5, i.e. the deterioration degree index S, evaluating the marking surface condition according to the marking surface condition evaluation rule, wherein the marking surface condition evaluation rule is, if the deterioration degree index S is less than 0.1, it indicates that the marking surface condition is good, if the deterioration degree index S is between 0.1 and 0.3, it indicates that the marking surface condition is general, if the deterioration degree index S is between 0.3 and 0.5, it indicates that the marking surface condition is poor, if the deterioration degree index S is greater than or equal to 0.5, it indicates that the marking surface condition is severely deteriorated; Step 7, storing the detection data and the deterioration degree index result, and formulating marking maintenance suggestions according to the marking surface condition corresponding to the deterioration degree index, in addition, forming a dynamic evolution curve of marking deterioration according to the deterioration degree indexes obtained at different times, and providing a basis for marking maintenance decision.

2. The method for automated assessment of road marking surface defects according to claim 1, characterized in that, In step 1, the camera device is an industrial camera with a resolution of not less than 1920x1080 pixels, a CMOS image sensor with a pixel count of not less than 2 million; the camera device is fixedly installed on the roof of the detection vehicle, the installation height from the roof is 1-1.5 meters, the camera angle is adjusted to 30°-45° according to the width of the detection road, and when the image is collected, the vehicle speed is controlled at 60-80 km / h; for the selected collection marking line, a static picture is obtained by intercepting a frame picture from a dynamic video, and each picture contains clear marking line with a length of 4-6 meters.

3. The method for automated assessment of pavement marking surface defects according to claim 1, wherein, In step 2, first, a two-dimensional Gaussian filter algorithm is used to smooth the image, effectively suppressing noise while preserving edge details as much as possible; then, using local histogram equalization method, the image is divided into multiple local regions, the gray histogram of each region is calculated and equalized independently, the image contrast is enhanced, and the gray difference between the marking line and the background is highlighted; then, using an improved Retinex light compensation algorithm, the image is decomposed into reflectivity and illumination two parts, the illumination component is adjusted, the influence of environmental light changes is eliminated, and the image brightness uniformity and color stability are improved; Subsequently, the image is geometrically corrected and perspective corrected to eliminate the deformation caused by the camera angle and ensure the accuracy of the marking line shape; finally, the image resolution and format are uniformly adjusted to complete the format standardization.

4. The method for automated assessment of pavement marking surface defects according to claim 1, wherein, In step 3, for crack detection, first, an edge detection operator is applied to extract the crack edge and capture the part with obvious gray change in the image; a morphological thinning algorithm is used to thin the crack edge to a single-pixel width to improve the positioning accuracy; an area filtering method is used to remove small noise regions; Then, through connected region analysis, according to the ratio of the longitudinal length of the connected region to the image height, the through cracks are screened out to ensure that only the effective cracks are retained; finally, morphological dilation and erosion operations are used to connect the crack breakpoints to form a continuous and complete crack mask.

5. The method for automated assessment of road marking surface defects according to claim 4, characterized in that, In step 3, after the two-dimensional Gaussian filter smoothing is performed on the gray image to reduce noise interference, the Canny operator is used for edge detection, the threshold is set to 0.2 to 0.3, and the crack edges with significant gray scale change are captured; then the edge is refined to a single-pixel width through a morphological thinning algorithm, and small noise areas are removed by applying area filtering; combined with connected region analysis, through-screen cracks with a longitudinal length proportion not less than 0.8 of the image height are screened, and the real crack part is effectively reserved; then the morphological dilation and corrosion operation is used to connect the crack breakpoints, form a continuous and complete crack mask, and count the number of cracks .

6. The method for automated assessment of pavement marking surface defects according to claim 4, wherein, In step 3, for damage detection, first, edge detection is performed to identify the damage area boundary; morphological dilation and closing operation is applied to fill the gap between the edges to realize the connection of the damage area; by calculating the connected region area, small areas smaller than a predetermined threshold are removed, and larger and more significant damage areas are retained; finally, a damage mask accurately representing the damage position is generated.

7. The method for automated assessment of road marking surface defects according to claim 6, characterized in that, In step 3, Canny edge detection is applied to the gray image with a threshold range of 0.1 to 0.3, and then dilation and closing operations are performed to fill the edge cracks with a structure element radius of 2 and 3 respectively; through connected region area analysis, small areas smaller than 1500 pixels are removed, only significant damage areas are retained, and a mask accurately reflecting the damage position is generated, which surrounds the significant damage area with a curve in the marking picture; According to the proportion of the number of pixels in the damage mask to the number of pixels in the entire image, the damage area proportion is calculated , wherein represents the damage area proportion, represents the number of pixels marked as damaged in the damage mask, represents the total number of pixels in the image.

8. The method for automated assessment of road marking surface defects according to claim 6, characterized in that, In step 3, for pollution detection, first, the preprocessed image is converted to HSV color space, and the saturation and brightness channels are extracted; Based on the preset threshold conditions, low saturation, low brightness and low gray pixel regions are screened out to preliminarily determine the pollution candidate area; Then, the pollution candidate region is logically "NOT" operated with the damage mask to eliminate the interference of the damage region to the pollution detection. Finally, the pure pollution mask is obtained for the subsequent quantitative analysis of the pollution degree.

9. The method for automated assessment of road marking surface defects according to claim 8, characterized in that, In step 3, the color image is first converted into HSV color space, the saturation and brightness channels are extracted, the saturation threshold is set to 0.08, the brightness threshold is set to 0.97, and the grayscale threshold is set to 245; the pixel region satisfying the conditions of saturation lower than 0.08, brightness lower than 0.97, and grayscale lower than 245 is preliminarily determined as a pollution candidate area; in order to avoid interference of a damaged area, a logical "not" operation is used to remove the area covered by a damaged mask to obtain a pure pollution mask, then the average grayscale, average saturation, and average brightness of the pollution area are calculated, and the saturation pollution index , the brightness pollution index , and the grayscale pollution index are calculated respectively; Saturation Pollution Index The saturation pollution index is calculated as follows: wherein, is the average saturation in the pollution mask region; the lower the saturation, the closer to 1, the more severe the pollution; the higher the saturation, the closer this index is to 0, indicating little or no pollution. Brightness pollution index For measuring the decrease of average brightness of a polluted area relative to normal brightness, the calculation method is: Wherein, is the average brightness of the polluted area; when the brightness is lower, the value is larger, indicating that pollution causes the surface to be dim; when the brightness is higher, the value is lower, indicating that the pollution has less impact; Gray pollution index Reflecting the relative position of the gray level of the pollution area relative to the clean area and the severely polluted area, the calculation method is: , wherein, is the average gray value of the pollution mask area, C gray is the maximum gray value of the clean mark area, indicating the upper limit of the gray level of the normal non-polluted mark area; P gray is the minimum gray value of the pollution area, indicating the lower limit of the gray level of the severely polluted part; the gray pollution index is normalized to 0 to 1, The lower the value, the more serious the pollution. After that, a pollution degree index L that reflects the degree of influence of pollution on the reticle surface is calculated pollution The calculation method is as follows: wherein, represents a pollution area ratio that represents the size of the pollution range, and the calculation method is as follows: , represents the number of pixels marked as pollution in the pollution mask, represents the total number of pixels of the image.

10. The method for automated assessment of pavement marking surface defects according to claim 8, wherein, In step 4, the crack number and the damage area ratio are normalized to convert the two indicators to the same range of 0 to 1; first, the crack number normalization index is calculated , wherein, is the set maximum number of through cracks, and more than this value is considered to be serious cracking of the marking; the setting principle of the maximum number of through cracks is based on the road grade; for expressways and national and provincial highways, 3 or more through cracks are considered to be serious cracking; for low-grade roads, 5 or more through cracks are considered to be serious cracking; Then the damage area proportion normalized index is calculated , wherein is the damage rate threshold value, and a value exceeding this value is considered as a serious damage to the reticle.

11. The method for automated assessment of pavement marking surface defects according to claim 10, wherein, is 0.

03.

12. The method for automated assessment of pavement marking surface defects of claim 10, wherein, In step 5, the crack number normalization index obtained in step 4 is , the damage area proportion normalization index , and the pollution degree index L pollution The degradation degree index S of the road marking surface is calculated by weighted summation according to the preset weights, and the formula is as follows: wherein the preset weights are respectively denoted by , , with the sum of the weights being 1.

13. The method for automated assessment of pavement marking surface defects according to claim 12, wherein, In step 5, , , .

14. The method for automated assessment of pavement marking surface defects of claim 12, wherein, In step 7, the dynamic evolution curve of the marking deterioration trend and the proposal of the maintenance suggestion are realized by the following steps: S701, the image data of the road marking is collected on the fixed date of each month; S702, the deterioration degree index S of the marking surface is obtained according to steps 2 to 6; S703, the deterioration degree index S obtained each month is composed into a time sequence to form the dynamic evolution curve of the marking deterioration; S704, the time sequence analysis method is applied to eliminate the influence of short-term fluctuations and judge the rising, stable or falling trend of the deterioration degree index S; S705, the maintenance suggestion is proposed according to the deterioration degree index S and the change trend thereof: The deterioration degree index S is low and shows a stable or falling trend, the maintenance suggestion is regular patrol, data monitoring and keeping the status quo; The deterioration degree index S is medium but continuously rises, the maintenance suggestion is timely maintenance including damage repair and surface coating; The deterioration degree index S is high and shows a stable or rising trend, the maintenance suggestion is emergency maintenance.

15. The method for automated assessment of pavement marking surface defects according to claim 14, wherein, In step 7, the time sequence analysis method is a moving average or a trend line fitting or a regression analysis.

Images