Road maintenance patrol system and method based on artificial intelligence

Abstract

The invention relates to the technical field of image processing, and discloses a road maintenance patrol system and method based on artificial intelligence, and the system comprises a collection unit which collects the road information of a to-be-patrolled area and determines the patrol speed and the patrol interval duration; the processing unit collects image data of the to-be-patrolled area at the patrolling speed, carries out gray processing on the image data and segments the image data; a judgment unit collects a pixel gray value in each piece of sub-image data and judges whether abnormal pixel points exist or not; comparing the sub-image data with abnormal pixel points with standard image data to determine suspected defect segments; the recognition unit performs feature recognition on the suspected defect segment based on image recognition and determines a defect score according to all defect features; and the adjusting unit determines a next maintenance patrol strategy according to the defect score. According to judgment and screening of the suspected defect segments and dynamic adjustment of the maintenance patrol strategy, the automation level of road patrol is improved, and the operation cost is reduced.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to an artificial intelligence-based road maintenance inspection system and method. Background Art

[0002] With the acceleration of urbanization, the pressure on urban roads continues to increase, and road maintenance and repair have become increasingly important. Traditional road inspection and maintenance work relies on manual on-site inspection and evaluation. Manual inspections are inefficient, labor-intensive, and easily limited by external factors such as weather and light, resulting in inaccurate and delayed inspection data. In addition, manual inspections are highly subjective, and the evaluation results are affected by the experience and judgment of the inspectors, lacking consistency and accuracy.

[0003] In recent years, with the rapid development of artificial intelligence, intelligent inspection systems based on artificial intelligence have gradually become a trend in modern road maintenance. Using drones, automated cameras and sensors, combined with image processing and analysis technology, road conditions can be monitored efficiently and accurately without relying on human intervention. However, when performing defect detection on roads based on artificial intelligence, the inspection system usually relies on preset inspection speeds and time intervals, and does not dynamically adjust according to the actual conditions of different roads, resulting in low inspection efficiency or serious missed inspections. In addition, when performing recognition, it is usually necessary to analyze and process each pixel of the image and recognize frame by frame. The large workload of recognition places high demands on the equipment, increasing operating costs.

[0004] Therefore, it is necessary to design a road maintenance inspection system and method based on artificial intelligence to solve the problems existing in current technology. Summary of the invention

[0005] In view of this, the present invention proposes an artificial intelligence-based road maintenance inspection system and method, aiming to solve the problems existing in the current artificial intelligence-based road maintenance inspection technology, such as lack of dynamic adjustment capability, large image recognition workload, time-consuming recognition process and high operating cost.

[0006] In one aspect, the present invention provides a road maintenance inspection system based on artificial intelligence, comprising: A collection unit is configured to collect road information of the area to be inspected, and determine the inspection speed and inspection interval duration according to the road information, wherein the road information includes road length, road width, and average road light ratio; a processing unit configured to collect image data of the area to be inspected at the inspection speed, perform grayscale processing on the image data, and segment the image data according to light and shadow to obtain a plurality of sub-image data; A judgment unit is configured to collect the pixel grayscale value in each of the sub-image data, and judge whether there is an abnormal pixel point according to the pixel grayscale value; compare the sub-image data with the abnormal pixel point with the standard image data to determine a similarity value, and when the similarity value is not within a normal similarity interval, regard the sub-image data as a suspected defect segment; an identification unit configured to perform feature identification on the suspected defective segment based on image recognition, extract defect features when defects exist in the sub-image data, and determine a defect score based on all defect features; The adjustment unit is configured to determine the next maintenance inspection strategy according to the defect score.

[0007] Furthermore, when the acquisition unit determines the inspection speed and the inspection interval according to the road information, it includes:

[0008]

[0009] Among them, V represents the patrol speed, L represents the road length, W represents the road width, I represents the average road light ratio, T represents the patrol interval, and T is rounded up, F represents the influence coefficient of road surface attachments, and F ranges from [0.05 to 0.2], and T0 represents the basic patrol interval.

[0010] Furthermore, the processing unit performs grayscale processing on the image data and divides the image data according to light and shadow to obtain a plurality of sub-image data, including: The processing unit obtains a grayscale image by performing weighted averaging on the image data; Taking the grayscale values ​​of each column of pixels in the grayscale image as a whole, obtaining a grayscale mean; When the difference between the grayscale mean and the grayscale mean of the previous column is greater than a difference threshold, the grayscale image is segmented.

[0011] Furthermore, when the judging unit judges whether there is an abnormal pixel point according to the pixel gray value, it includes: Determine any point in the sub-image data as the origin; The verification radius is determined with the origin as the center and r as the neighborhood radius; Calculate the verification average pixel grayscale value according to the pixel grayscale value of each coordinate point within the verification radius; When the difference between the pixel grayscale value of the origin and the verified average pixel grayscale value is greater than the difference threshold, the origin is determined to be an abnormal pixel; When there are abnormal pixels in the sub-image data, the judgment unit compares the sub-image data with the standard image data to determine a similarity value; When there are no abnormal pixels in the sub-image data, the judgment unit regards the sub-image data as normal image data.

[0012] Furthermore, when the judgment unit compares the sub-image data with the standard image data to determine the similarity value, it includes: The judging unit extracts standard sub-image data at the same position as the sub-image data from the standard image data, compares the sub-image data with the standard sub-image data, and obtains the similarity between the sub-image data and the standard sub-image data; The judgment unit compares the similarity with a normal similarity interval. When the similarity value is within the normal similarity interval, the judgment unit uses the sub-image data as the normal image data; when the similarity value is not within the normal similarity interval, the judgment unit uses the sub-image data as the suspected defect segment.

[0013] Furthermore, when the judgment unit obtains the normal similarity interval, it includes: The judgment unit collects historical normal image data, and filters the historical normal sub-image data corresponding to the position of the sub-image data in the historical normal image data, obtains the similarity values ​​of all the historical normal sub-image data and the standard sub-image data and integrates them into a similarity set, determines the maximum similarity value and the minimum similarity value in the similarity set, and determines the normal similarity interval according to the maximum similarity value and the minimum similarity value, and the normal similarity interval includes a left boundary and a right boundary.

[0014] Furthermore, when the identification unit performs feature identification on the suspected defective segment based on image recognition, it includes: When there is no defect in the suspected defect segment, the identification unit uses the suspected defect segment as the normal image data; When there is a defect in the suspected defect segment, the identification unit extracts defect features, wherein the defect features include cracks, pits, and peeling; When the identification unit determines the defect score according to all defect features, it includes: The identification unit obtains a defect ratio of the defect area to the total road area according to the defect characteristics, and determines a defect score according to the defect ratio, wherein the defect score is proportional to the defect ratio.

[0015] Furthermore, when the adjustment unit determines the next maintenance inspection strategy according to the defect score, it includes: The adjustment unit compares the defect score with the score threshold, and determines the next maintenance inspection strategy according to the comparison result; When the defect score is less than or equal to the score threshold, the adjustment unit continues to operate at the inspection speed and inspection interval duration; When the defect score is greater than the score threshold, the adjustment unit adjusts the inspection interval duration according to the defect score and operates with the adjusted inspection interval duration.

[0016] Further, when the adjustment unit adjusts the inspection interval duration according to the defect score, it includes: The adjustment unit obtains a score difference according to the defect score and the score threshold, the score difference being the difference between the defect score and the score threshold, the adjustment unit compares the score difference with the first preset difference and the second preset difference respectively, and determines an adjustment coefficient according to the comparison result to adjust the inspection interval duration, wherein the adjustment coefficient has a value range of [0.5, 0.9]; When the score difference is less than or equal to the first preset difference, the first adjustment coefficient is determined to adjust the inspection interval duration; when the score difference is greater than the first preset difference and less than or equal to the second preset difference, the second adjustment coefficient is determined to adjust the inspection interval duration; when the score difference is greater than the second preset difference, the third adjustment coefficient is determined to adjust the inspection interval duration; the first adjustment coefficient is greater than the second adjustment coefficient, and the second adjustment coefficient is greater than the third adjustment coefficient.

[0017] Compared with the prior art, the beneficial effects of the present invention are: dynamically adjusting the inspection speed and interval duration according to the specific information of the road, improving the inspection efficiency and accuracy, analyzing the overall image through grayscale processing and light and shadow segmentation technology, facilitating subsequent detection and judgment, and avoiding the inefficiency and omissions of traditional manual inspections. The judgment unit screens suspected defect fragments based on abnormal pixel points and similarity comparison judgments, avoiding frame-by-frame analysis when performing defect identification, and reducing image recognition pressure. Through the deep identification and feature extraction of suspected defect fragments combined with defect scoring to dynamically adjust the maintenance inspection strategy, the automation, accuracy and intelligence level of road inspections are improved, and the operating costs are reduced.

[0018] On the other hand, the present application also provides an artificial intelligence-based road maintenance inspection method, which is applied to the above-mentioned artificial intelligence-based road maintenance inspection system, including: Collecting road information of the area to be inspected, and determining the inspection speed and inspection interval duration according to the road information, wherein the road information includes road length, road width, and average road light ratio; Collecting image data of the area to be inspected at the inspection speed, performing grayscale processing on the image data, and segmenting the image data according to light and shadow to obtain a plurality of sub-image data; Collecting the pixel grayscale value in each of the sub-image data, and judging whether there is an abnormal pixel point according to the pixel grayscale value; comparing the sub-image data with the abnormal pixel point with the standard image data to determine the similarity value, and when the similarity value is not within the normal similarity interval, treating the sub-image data as a suspected defect fragment; Performing feature recognition on the suspected defective segment based on image recognition, extracting defect features when defects exist in the sub-image data, and determining a defect score based on all defect features; The next maintenance inspection strategy is determined based on the defect score.

[0019] It is understandable that the above-mentioned artificial intelligence-based road maintenance inspection system and method have the same beneficial effects and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings: Figure 1 A structural block diagram of a road maintenance inspection system based on artificial intelligence provided by an embodiment of the present invention; Figure 2 A flowchart of an artificial intelligence-based road maintenance inspection method provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0021] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art. It should be noted that, in the absence of conflict, the embodiments of the present invention and the features described in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0022] In some embodiments of the present application, see Figure 1 As shown, a road maintenance inspection system based on artificial intelligence includes: The collection unit is configured to collect road information of the area to be inspected, and determine the inspection speed and inspection interval duration based on the road information. The road information includes road length, road width, and average road light ratio.

[0023] The processing unit is configured to collect image data of the area to be inspected at an inspection speed, perform grayscale processing on the image data, and segment the image data according to light and shadow to obtain a plurality of sub-image data.

[0024] The judgment unit is configured to collect the pixel grayscale value in each sub-image data, and judge whether there is an abnormal pixel point according to the pixel grayscale value. The sub-image data with abnormal pixels is compared with the standard image data to determine the similarity value, and when the similarity value is not within the normal similarity range, the sub-image data is regarded as a suspected defect segment.

[0025] The recognition unit is configured to perform feature recognition on the suspected defective segment based on image recognition, extract defect features when defects exist in the sub-image data, and determine a defect score based on all defect features.

[0026] The adjustment unit is configured to determine the next maintenance inspection strategy according to the defect score.

[0027] Specifically, the acquisition unit obtains the road information of the area to be inspected. The road information includes the length, width and average light ratio of the road. These data will be used to calculate the speed and interval of the inspection, so as to dynamically adjust the inspection strategy to meet the needs of different road conditions. The processing unit is responsible for the acquisition and preprocessing of image data. The inspection system collects image data on the road according to the set inspection speed and performs grayscale processing on the image. Grayscale processing is an image preprocessing method that reduces the complexity of data by converting color images into grayscale images, which is convenient for subsequent processing. Converting the original image into a grayscale image, removing color information and retaining only brightness information can reduce the computational complexity and highlight the structural features in the image. The image area is segmented by analyzing the light and shadow changes in the image. Light and shadow changes can help identify different image areas, especially can highlight defects such as cracks and potholes on the road surface. After image segmentation, the overall image is segmented into several sub-image data to facilitate more detailed defect detection and analysis of each area. The judgment unit judges whether there are abnormal pixels in the image by the pixel grayscale value. The pixel grayscale value of each sub-image data will be analyzed, and by comparing the fluctuation of the grayscale value, it is determined whether there is an abnormality. If the grayscale value of a certain area fluctuates abnormally, it indicates that there may be a defect. The sub-image data of the suspected defect area is compared with the standard image data to calculate the similarity value. If the similarity value is not within the normal similarity range, the sub-image data is marked as a "suspected defect fragment". By comparing, it is determined whether there is a significant defect area in the image, so as to screen out the part that needs further identification and processing. The recognition unit performs feature recognition on the suspected defect fragment. Different types of defects such as cracks, pits, peeling, etc. are identified through image recognition algorithms. For suspected defect fragments, their features such as shape, size, depth, etc. are extracted through image recognition technology to distinguish different defect types. According to the pre-trained model, the specific defect type in the image is identified and the severity of the defect is determined. Automatically classify and mark the defect area to reduce the need for manual intervention. After the defect is identified, the adjustment unit will adjust the next inspection strategy according to the severity score of the defect. The defect score is obtained through a comprehensive analysis of multiple parameters such as the area, depth, and length of the defect. Scoring is based on the size, type, scope and severity of the defect. The higher the score, the more serious the defect. According to the defect score, the inspection strategy is dynamically adjusted. For example, if there are more serious defects in certain areas, the inspection frequency of the area is increased.

[0028] It is understandable that by combining artificial intelligence, computer vision and dynamic adjustment mechanisms, fully automatic and intelligent road maintenance inspections are achieved. The inspection speed and interval duration are dynamically adjusted based on the actual information of the road, which improves the efficiency and accuracy of the inspection and avoids the inefficiency and omissions of traditional methods. Image processing technology (including grayscale processing and light and shadow segmentation) makes the processing of image data easier and more efficient, providing a clear basis for subsequent defect detection. The judgment unit screens out suspected defect areas through pixel grayscale analysis and similarity comparison technology, avoiding the calculation pressure of frame-by-frame analysis and reducing operating costs. Combined with image recognition technology for defect identification and scoring, the maintenance inspection strategy is automatically adjusted, realizing dynamic adjustment of the inspection strategy according to the severity of the defect.

[0029] In some embodiments of the present application, when the acquisition unit determines the patrol speed and the patrol interval duration according to the road information, it includes:

[0030]

[0031] Among them, V represents the patrol speed, L represents the road length, W represents the road width, I represents the average road light ratio, T represents the patrol interval, and T is rounded up, F represents the influence coefficient of road surface attachments, and F ranges from [0.05 to 0.2], and T0 represents the basic patrol interval.

[0032] Specifically, for example: the road length is 5000 meters, the road width is 12 meters, the average light ratio is 0.75, and the patrol speed after calculation is 3.6km / h. The influence coefficient of road surface attachments is 0.1, T0 is 48h, and the final patrol interval is about 133h.

[0033] It is understandable that by normalizing road information parameters (length, width, and light ratio) to quantify and standardize the impact of different road conditions on patrol speed and interval duration, a more intelligent and personalized patrol scheduling can be achieved. Dynamically adjust the patrol speed and interval duration, optimize the patrol process according to actual road conditions, improve patrol efficiency, and ensure that road maintenance work is carried out efficiently and accurately. The introduction of the attachment influence coefficient F can compensate for the patrol frequency and quality according to environmental conditions, making the system more robust.

[0034] In some embodiments of the present application, when the processing unit performs grayscale processing on the image data and divides the image data according to light and shadow to obtain a plurality of sub-image data, it includes: The processing unit obtains a grayscale image by performing weighted averaging on the image data.

[0035] Take the grayscale value of each column of pixels in the grayscale image as a whole and obtain the grayscale mean.

[0036] When the difference between the grayscale mean and the grayscale mean of the previous column is greater than the difference threshold, the grayscale image is segmented.

[0037] It can be understood that the processing unit obtains a grayscale image by weighted averaging the image data. Grayscale is a common step in image processing, which converts a color image into a single grayscale value representation. The grayscale value of each pixel usually represents the brightness or grayscale level of the pixel, and this processing simplifies the image information, making subsequent processing (such as segmentation, feature extraction, etc.) more efficient. Weighted average method: For the color of each pixel, the values ​​of the RGB (red, green, and blue) three-color channels are converted into grayscale values ​​by weighted averaging. The processing unit obtains the overall brightness information of each column by calculating the grayscale mean of each column in the grayscale image. In order to determine which areas in the image have undergone significant light and shadow changes, the difference between the grayscale means of two adjacent columns is calculated. If the grayscale mean difference between the two columns is greater than the preset difference threshold, it is considered that there is a significant light and shadow change between the two columns, and the image is segmented here. According to the above grayscale mean difference, the image is segmented into multiple sub-image data (i.e., different areas of the image). Each sub-image data represents a "paragraph" of the image.

[0038] It can be understood that by using the weighted average method for grayscale processing and combining it with the segmentation technology based on light and shadow changes, the image data processing process is simplified and the efficiency and accuracy of image analysis are improved. Grayscale processing focuses the image information on brightness changes, while the segmentation strategy accurately identifies the significant change areas in the image through the difference in grayscale mean values, thereby effectively segmenting the sub-image data related to road defects. This reduces the computational complexity, avoids unnecessary calculations, and ensures the accuracy of image segmentation.

[0039] In some embodiments of the present application, when the judgment unit judges whether there is an abnormal pixel point according to the pixel gray value, it includes: A point in the sub-image data is arbitrarily determined as the origin.

[0040] The verification radius is determined with the origin as the center and r as the neighborhood radius.

[0041] The verification average pixel grayscale value is calculated based on the pixel grayscale value of each coordinate point within the verification radius.

[0042] When the difference between the pixel grayscale value of the origin and the verified average pixel grayscale value is greater than the difference threshold, the origin is determined to be an abnormal pixel.

[0043] When there are abnormal pixels in the sub-image data, the judgment unit compares the sub-image data with the standard image data to determine a similarity value.

[0044] When there are no abnormal pixels in the sub-image data, the judging unit regards the sub-image data as normal image data.

[0045] It can be understood that by judging abnormal pixels based on the difference in local pixel grayscale values, the high computational complexity of pixel-by-pixel processing of the entire image is effectively avoided, and the detection accuracy of subtle defects is high. Through the dynamic adjustment of the neighborhood radius and the setting of the difference threshold, it can flexibly adapt to different types of road images and their defects, reducing the probability of false detection and missed detection.

[0046] In some embodiments of the present application, when the judgment unit compares the sub-image data with the standard image data to determine the similarity value, it includes: the judgment unit extracts the standard sub-image data at the same position as the sub-image data in the standard image data, compares the sub-image data with the standard sub-image data, and obtains the similarity between the sub-image data and the standard sub-image data. The judgment unit compares the similarity with the normal similarity interval, and when the similarity value is within the normal similarity interval, the judgment unit regards the sub-image data as normal image data. When the similarity value is not within the normal similarity interval, the judgment unit regards the sub-image data as a suspected defective fragment.

[0047] In some embodiments of the present application, when the judgment unit obtains the normal similarity interval, it includes: the judgment unit collects historical normal image data, and filters the historical normal sub-image data corresponding to the sub-image data position in the historical normal image data, obtains the similarity values ​​of all historical normal sub-image data and the standard sub-image data and integrates them into a similarity set, determines the maximum similarity value and the minimum similarity value in the similarity set, and determines the normal similarity interval according to the maximum similarity value and the minimum similarity value, and the normal similarity interval includes a left boundary and a right boundary.

[0048] It is understandable that after extracting the standard sub-image data, the judgment unit compares the sub-image data with the standard sub-image data and calculates the similarity between them. The similarity calculation can use the mean square error or the structural similarity index. By calculating the similarity between images, it is determined whether the image data contains defects. By obtaining the similarity interval from the historical normal image data and comparing it with the standard image data, the image area that meets the normal state can be effectively screened out, and the area where defects may exist can be marked as suspected defects. It is better to eliminate errors caused by external factors such as lighting and shooting angles, and improve the accuracy of defect detection. The similarity calculation method based on historical data and standard images has higher accuracy and adaptability, and can effectively process road image data under different conditions. At the same time, by dynamically adjusting the normal similarity interval, it can flexibly adapt to different types of road surfaces according to actual data conditions, reduce the risk of misjudgment and missed judgment, and improve the intelligence and precision level of the road inspection system.

[0049] In some embodiments of the present application, when the recognition unit performs feature recognition on the suspected defective fragment based on image recognition, it includes: When there is no defect in the suspected defect segment, the identification unit regards the suspected defect segment as normal image data.

[0050] When there are defects in the suspected defective segment, the recognition unit extracts defect features, which include cracks, pits, and peeling.

[0051] The recognition unit determines the defect score based on all defect characteristics, including: The recognition unit obtains the defect ratio of the defect area to the total road area according to the defect characteristics, and determines the defect score according to the defect ratio. The defect score is proportional to the defect ratio.

[0052] It is understandable that edge detection can be used to identify defects such as cracks, potholes and peeling in the image when performing defect detection. If a defect is detected, the recognition unit will extract the characteristics of the defect, which include: Cracks: elongated cracks or splits appear on the road surface. Potholes: larger road surface damage, forming sunken holes. Peeling: the shedding of road surface material. By analyzing the characteristics of the defects, the defect fragments are scored to determine the degree of defects in the entire road area. The basis for scoring is mainly the proportion of defects, that is, the proportion of the defect area to the total area of ​​the road. By introducing the determination of defect scores, a quantitative assessment of road defects is achieved. Compared with traditional methods, automated feature recognition and scoring mechanisms improve the accuracy and efficiency of road maintenance inspections and avoid subjective errors and omissions in manual evaluations. In addition, the application of defect proportions can dynamically adjust the maintenance priority according to the actual degree of damage, making the inspection results more targeted and reliable, ensuring that maintenance work can be carried out accurately and efficiently, saving resources and extending the service life of the road.

[0053] In some embodiments of the present application, when the adjustment unit determines the next maintenance inspection strategy based on the defect score, it includes: the adjustment unit compares the defect score with the score threshold, and determines the next maintenance inspection strategy based on the comparison result.

[0054] When the defect score is less than or equal to the score threshold, the adjustment unit continues to run at the inspection speed and inspection interval. When the defect score is greater than the score threshold, the adjustment unit adjusts the inspection interval according to the defect score and runs at the adjusted inspection interval.

[0055] In some embodiments of the present application, when the adjustment unit adjusts the inspection interval duration according to the defect score, it includes: the adjustment unit obtains a score difference according to the defect score and the score threshold, the score difference is the difference between the defect score and the score threshold, the adjustment unit compares the score difference with the first preset difference and the second preset difference respectively, determines the adjustment coefficient according to the comparison result to adjust the inspection interval duration, and the adjustment coefficient takes a value range of [0.5, 0.9].

[0056] Specifically, when the score difference is less than or equal to the first preset difference, the first adjustment coefficient is determined to adjust the inspection interval. When the score difference is greater than the first preset difference and less than or equal to the second preset difference, the second adjustment coefficient is determined to adjust the inspection interval. When the score difference is greater than the second preset difference, the third adjustment coefficient is determined to adjust the inspection interval. The first adjustment coefficient is greater than the second adjustment coefficient, and the second adjustment coefficient is greater than the third adjustment coefficient.

[0057] It is understandable that by introducing a dynamic patrol interval adjustment mechanism, a more intelligent road maintenance patrol is achieved. According to the different road defect scores, it can automatically decide whether to increase the patrol frequency, thus avoiding the inefficient patrol arrangements in traditional manual patrols. Especially when the road is severely damaged, by reducing the patrol interval time, it is ensured that new defects can be discovered and handled in time. The patrol efficiency and accuracy are improved, and patrol resources are reasonably allocated according to the actual situation of the road, reducing the unnecessary costs caused by excessive patrols.

[0058] In the above embodiment, the inspection speed and interval time are dynamically adjusted according to the specific information of the road, which improves the inspection efficiency and accuracy. The overall image is analyzed through grayscale processing and light and shadow segmentation technology, which is convenient for subsequent detection and judgment, avoiding the inefficiency and omissions of traditional manual inspections. The judgment unit screens suspected defect fragments based on abnormal pixel points and similarity comparison judgments, avoiding frame-by-frame analysis when performing defect identification, and reducing image recognition pressure. Through the deep identification and feature extraction of suspected defect fragments combined with defect scoring to dynamically adjust the maintenance inspection strategy, the automation, accuracy and intelligence level of road inspections are improved, and the operating cost is reduced.

[0059] In another preferred embodiment based on the above embodiment, refer to Figure 2 As shown, this embodiment provides a road maintenance inspection method based on artificial intelligence, which is applied to the above-mentioned road maintenance inspection system based on artificial intelligence, including: S100: Collecting road information of the area to be inspected, and determining the inspection speed and inspection interval according to the road information, wherein the road information includes road length, road width, and average road light ratio; S200: collecting image data of the area to be inspected at an inspection speed, performing grayscale processing on the image data, and segmenting the image data according to light and shadow to obtain a plurality of sub-image data; S300: Collecting the pixel grayscale value in each sub-image data, judging whether there is an abnormal pixel point according to the pixel grayscale value; comparing the sub-image data with the abnormal pixel point with the standard image data to determine the similarity value, and when the similarity value is not within the normal similarity range, treating the sub-image data as a suspected defect segment; S400: Perform feature recognition on the suspected defective segment based on image recognition, extract defect features when defects exist in the sub-image data, and determine a defect score based on all defect features.

[0060] S500: Determine the next maintenance inspection strategy based on the defect score.

[0061] It is understandable that the inspection speed and interval time are adjusted dynamically according to the specific information of the road, which improves the inspection efficiency and accuracy. The overall image is analyzed through grayscale processing and light and shadow segmentation technology, which is convenient for subsequent detection and judgment, avoiding the inefficiency and omissions of traditional manual inspections. The judgment unit screens suspected defect fragments based on abnormal pixels and similarity comparison judgments, avoiding frame-by-frame analysis when identifying defects and reducing image recognition pressure. Through the deep recognition and feature extraction of suspected defect fragments combined with defect scoring to dynamically adjust the maintenance inspection strategy, the automation, accuracy and intelligence level of road inspections are improved, and the operating costs are reduced.

[0062] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0063] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0064] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0065] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0066] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A road maintenance inspection system based on artificial intelligence, characterized in that: include: A collection unit is configured to collect road information of the area to be inspected, and determine the inspection speed and inspection interval duration according to the road information, wherein the road information includes road length, road width, and average road light ratio; a processing unit configured to collect image data of the area to be inspected at the inspection speed, perform grayscale processing on the image data, and segment the image data according to light and shadow to obtain a plurality of sub-image data; A judging unit configured to collect pixel grayscale values ​​in each of the sub-image data, and judge whether there is an abnormal pixel point according to the pixel grayscale value; Compare the sub-image data with abnormal pixels with the standard image data to determine a similarity value, and when the similarity value is not within a normal similarity interval, regard the sub-image data as a suspected defect segment; an identification unit configured to perform feature identification on the suspected defective segment based on image recognition, extract defect features when defects exist in the sub-image data, and determine a defect score based on all defect features; The adjustment unit is configured to determine the next maintenance inspection strategy according to the defect score.

2. The road maintenance inspection system based on artificial intelligence according to claim 1 is characterized in that: When the acquisition unit determines the inspection speed and the inspection interval duration according to the road information, it includes: Among them, V represents the patrol speed, L represents the road length, W represents the road width, I represents the average road light ratio, T represents the patrol interval, and T is rounded up, F represents the influence coefficient of road surface attachments, and F ranges from [0.05 to 0.2], and T0 represents the basic patrol interval.

3. The road maintenance inspection system based on artificial intelligence according to claim 1 is characterized in that: When the processing unit performs grayscale processing on the image data and divides the image data according to light and shadow to obtain a plurality of sub-image data, it includes: The processing unit obtains a grayscale image by performing weighted averaging on the image data; Taking the grayscale values ​​of each column of pixels in the grayscale image as a whole, obtaining a grayscale mean; When the difference between the grayscale mean and the grayscale mean of the previous column is greater than a difference threshold, the grayscale image is segmented.

4. The road maintenance inspection system based on artificial intelligence according to claim 3 is characterized in that: When the judging unit judges whether there is an abnormal pixel point according to the pixel gray value, it includes: Determine any point in the sub-image data as the origin; The verification radius is determined with the origin as the center and r as the neighborhood radius; Calculate the verification average pixel grayscale value according to the pixel grayscale value of each coordinate point within the verification radius; When the difference between the pixel grayscale value of the origin and the verified average pixel grayscale value is greater than the difference threshold, the origin is determined to be an abnormal pixel; When there are abnormal pixels in the sub-image data, the judgment unit compares the sub-image data with the standard image data to determine a similarity value; When there are no abnormal pixels in the sub-image data, the judgment unit regards the sub-image data as normal image data.

5. The artificial intelligence-based road maintenance inspection system according to claim 4 is characterized in that: When the judgment unit compares the sub-image data with the standard image data to determine the similarity value, it includes: The judging unit extracts standard sub-image data at the same position as the sub-image data from the standard image data, compares the sub-image data with the standard sub-image data, and obtains the similarity between the sub-image data and the standard sub-image data; The judgment unit compares the similarity with a normal similarity interval. When the similarity value is within the normal similarity interval, the judgment unit uses the sub-image data as the normal image data; when the similarity value is not within the normal similarity interval, the judgment unit uses the sub-image data as the suspected defect segment.

6. The artificial intelligence-based road maintenance inspection system according to claim 5 is characterized in that: When the judging unit obtains the normal similarity interval, it includes: The judgment unit collects historical normal image data, and filters the historical normal sub-image data corresponding to the position of the sub-image data in the historical normal image data, obtains the similarity values ​​of all the historical normal sub-image data and the standard sub-image data and integrates them into a similarity set, determines the maximum similarity value and the minimum similarity value in the similarity set, and determines the normal similarity interval according to the maximum similarity value and the minimum similarity value, and the normal similarity interval includes a left boundary and a right boundary.

7. The artificial intelligence-based road maintenance inspection system according to claim 6 is characterized in that: When the recognition unit performs feature recognition on the suspected defective segment based on image recognition, it includes: When there is no defect in the suspected defect segment, the identification unit uses the suspected defect segment as the normal image data; When there is a defect in the suspected defect segment, the identification unit extracts defect features, wherein the defect features include cracks, pits, and peeling; When the identification unit determines the defect score according to all defect features, it includes: The identification unit obtains a defect ratio of the defect area to the total road area according to the defect characteristics, and determines a defect score according to the defect ratio, wherein the defect score is proportional to the defect ratio.

8. The artificial intelligence-based road maintenance inspection system according to claim 7 is characterized in that: When the adjustment unit determines the next maintenance inspection strategy according to the defect score, it includes: The adjustment unit compares the defect score with the score threshold, and determines the next maintenance inspection strategy according to the comparison result; When the defect score is less than or equal to the score threshold, the adjustment unit continues to operate at the inspection speed and inspection interval duration; When the defect score is greater than the score threshold, the adjustment unit adjusts the inspection interval duration according to the defect score and operates with the adjusted inspection interval duration.

9. The artificial intelligence-based road maintenance inspection system according to claim 8, characterized in that: When the adjustment unit adjusts the inspection interval duration according to the defect score, it includes: The adjustment unit obtains a score difference according to the defect score and the score threshold, the score difference being the difference between the defect score and the score threshold, the adjustment unit compares the score difference with the first preset difference and the second preset difference respectively, and determines an adjustment coefficient according to the comparison result to adjust the inspection interval duration, wherein the adjustment coefficient has a value range of [0.5, 0.9]; When the score difference is less than or equal to the first preset difference, the first adjustment coefficient is determined to adjust the inspection interval duration; when the score difference is greater than the first preset difference and less than or equal to the second preset difference, the second adjustment coefficient is determined to adjust the inspection interval duration; when the score difference is greater than the second preset difference, the third adjustment coefficient is determined to adjust the inspection interval duration; the first adjustment coefficient is greater than the second adjustment coefficient, and the second adjustment coefficient is greater than the third adjustment coefficient.

10. A road maintenance inspection method based on artificial intelligence, applied to the road maintenance inspection system based on artificial intelligence as claimed in any one of claims 1 to 9, characterized in that: include: Collecting road information of the area to be inspected, and determining the inspection speed and inspection interval duration according to the road information, wherein the road information includes road length, road width, and average road light ratio; Collecting image data of the area to be inspected at the inspection speed, performing grayscale processing on the image data, and segmenting the image data according to light and shadow to obtain a plurality of sub-image data; Collecting the pixel grayscale value in each of the sub-image data, and judging whether there is an abnormal pixel point according to the pixel grayscale value; comparing the sub-image data with the abnormal pixel point with the standard image data to determine the similarity value, and when the similarity value is not within the normal similarity interval, treating the sub-image data as a suspected defect fragment; Performing feature recognition on the suspected defective segment based on image recognition, extracting defect features when defects exist in the sub-image data, and determining a defect score based on all defect features; The next maintenance inspection strategy is determined based on the defect score.

Images