Image feature-based asphalt pavement crack identification method and system

Abstract

The invention relates to the technical field of image recognition, and discloses an asphalt pavement crack recognition method and system based on image features, and the method comprises the steps: collecting the feature information of a to-be-monitored pavement, and determining the size information of a single image during image collection; performing region division on the to-be-monitored pavement according to the size information, acquiring initial image data for each region, obtaining the number of cracks in the initial image data, and judging whether to expand the image data size; when the number of the cracks is not unique, establishing a data set and determining image data volume in the data set according to the number of the cracks; when a data set is established, crack characteristics of each image data are extracted, and crack characteristics of a representation representative graph determination area are determined based on a kernel density function; and integrating the crack characteristics of all the regions, classifying the crack characteristics of the to-be-monitored pavement through a clustering algorithm, and determining the crack grade of the to-be-monitored pavement. The effectiveness of data acquisition is improved, and the automation level and accuracy of asphalt pavement crack detection are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image recognition, and in particular to an asphalt pavement crack recognition method and system based on image features. Background Art

[0002] With the acceleration of urbanization and the continuous growth of transportation demand, the quality and maintenance management of asphalt pavement are becoming more and more important. Pavement cracks are one of the common diseases of asphalt pavement, which not only affects driving safety, but also accelerates the aging and damage of the pavement. Therefore, timely and accurate identification and evaluation of pavement cracks is a key link in maintenance work.

[0003] At present, traditional pavement crack detection methods mainly rely on manual visual inspection and simple image analysis technology. These methods have the following major problems: Manual detection is limited by the experience and subjective judgment of the inspector, which often leads to missed detection or misjudgment, affecting the accuracy of crack assessment. Existing image processing technologies mostly focus on simple edge detection and template matching, which are difficult to adapt to complex pavement conditions, such as different lighting, humidity and color difference. These factors will interfere with the identification of cracks and reduce the reliability of detection. The current methods for collecting and analyzing pavement features often lack systematicity and cannot effectively integrate multiple feature information, resulting in insufficient or unbalanced data volume, affecting the model training effect.

[0004] Therefore, it is necessary to design an asphalt pavement crack recognition method and system based on image features to solve the problems existing in current technology. Summary of the invention

[0005] In view of this, the present invention proposes an asphalt pavement crack identification method and system based on image features, aiming to solve the problem that most of the current methods rely on manual detection, which has low detection efficiency and poor effect, and even has low detection accuracy by image processing.

[0006] In one aspect, the present invention provides a method for identifying cracks in an asphalt pavement based on image features, comprising:

[0007] Collect characteristic information of the road surface to be monitored, the characteristic information including road surface size, color difference between the road surface and the surrounding area, and road surface flatness, and determine the size information of a single image during image acquisition based on the characteristic information;

[0008] Divide the road surface to be monitored into regions according to the size information, collect initial image data for each region, obtain the number of cracks in the initial image data based on image recognition, and determine whether to expand the image data volume according to the number of cracks;

[0009] When the number of cracks is not unique, a data set is established and the amount of image data in the data set is determined according to the number of cracks; when it is determined that the data set is established, crack features of each image data are extracted, a representative representation graph is determined based on a kernel density function, and regional crack features are determined according to the representative representation graph;

[0010] All the regional crack characteristics are integrated, the crack characteristics of the road surface to be monitored are classified by a clustering algorithm, and the crack grade of the road surface to be monitored is determined.

[0011] Furthermore, when determining the size information of a single image during image acquisition according to the characteristic information, it includes: obtaining a characteristic index of the road surface to be monitored according to the characteristic information, and determining the size information of the single image according to the characteristic index, wherein the characteristic index is calculated by the following formula:

[0012]

[0013] Among them, T represents the characteristic index, L represents the actual length of the road surface to be monitored, D1 represents the color difference measurement between the road surface and the surrounding area, and D2 represents the flatness measurement of the road surface.

[0014] Furthermore, when determining the size information of the single image according to the characteristic index, it includes:

[0015] The characteristic index is compared with a first preset characteristic index and a second preset characteristic index respectively, and the size area of ​​the single image is determined according to the comparison result, wherein the first preset characteristic index is smaller than the second preset characteristic index;

[0016] When the characteristic index is less than or equal to the first preset characteristic index, the size area is determined to be the first preset area; when the characteristic index is greater than the first preset characteristic index and less than or equal to the second preset characteristic index, the size area is determined to be the second preset area; when the characteristic index is greater than the second preset characteristic index, the size area is determined to be the third preset area; the first preset area is smaller than the second preset area, and the second preset area is smaller than the third preset area.

[0017] Further, judging whether to expand the image data volume according to the number of cracks includes:

[0018] When the number of cracks in the initial image data is unique, it is determined not to expand the image data, and all the regional crack features are integrated, and the crack features of the road surface to be monitored are classified by a clustering algorithm to determine the crack grade of the road surface to be monitored;

[0019] When the number of cracks in the initial image data is not unique, the amount of expanded image data is determined.

[0020] Further, when determining to expand the image data volume, establishing a data set and determining the image data volume in the data set according to the number of cracks, it includes:

[0021]

[0022] Wherein, N represents the amount of image data, N is rounded up, C represents the constant factor, R represents the number of cracks, Q represents the light intensity for collecting the initial image data, and Q0 represents the average light intensity.

[0023] Furthermore, when determining the image data amount, establishing the data set also includes:

[0024] Image data is collected from different angles in the same area during the same period of time.

[0025] Furthermore, determining a representative image based on a kernel density function and determining regional crack characteristics according to the representative image includes:

[0026] Performing kernel density estimation on the data set to determine the image data with the highest probability as the representation representative image;

[0027] When performing the kernel density estimation, the Gaussian kernel density function is:

[0028]

[0029] Among them, n represents the total number of image data in the data set, h represents the smoothing bandwidth, and s i represents the crack characteristics of the i-th image data, and x represents any variable in the overall data.

[0030] Furthermore, the smoothing bandwidth is calculated by the following formula:

[0031]

[0032] Where h represents the smoothing bandwidth, σ represents the standard deviation of the crack characteristics of all image data in the data set, β1 represents the data skewness, β2 represents the data kurtosis, n represents the total number of image data in the data set, and k represents the adjustment coefficient;

[0033] The data skewness is calculated by the following formula:

[0034]

[0035] The data kurtosis is calculated by the following formula:

[0036]

[0037] Among them, n represents the total number of image data in the dataset, s irepresents the crack characteristics of the i-th image data, represents the mean of the crack characteristics of all image data, and σ represents the standard deviation of the crack characteristics of all image data.

[0038] Furthermore, the crack characteristics of the road surface to be monitored are classified by a clustering algorithm to determine the crack grade of the road surface to be monitored, including:

[0039] S1: Initialize K centroids in the classified historical road surface image data, and assign the crack features of the road surface to be monitored to the nearest centroid to form K clusters;

[0040] S2: Recalculate the centroid of each cluster;

[0041] S3: Repeat S1 and S2 until the center of mass no longer changes or the preset number of repetitions is completed;

[0042] S4: taking the classification and crack grade corresponding to the historical road surface image data in the cluster where the characterizing representative graph is located as the evaluation result of the crack characteristics of the road surface to be monitored;

[0043] The centroid calculation formula is as follows:

[0044]

[0045] Among them, Zk represents the centroid of the kth cluster, |Ck| represents the number of data in the kth cluster, and Ti represents the feature vector of the i-th data.

[0046] Compared with the prior art, the beneficial effects of the present invention are as follows: by comprehensively considering the road surface feature information (such as road surface size, color difference and flatness), determining the size information during image acquisition, and finely dividing the road surface to be monitored into regions, it is ensured that the cracks in each region are fully imaged. Based on image recognition technology, the number of cracks can be accurately obtained, and it is flexible to determine whether the image data volume needs to be expanded to cope with complex road conditions. When the number of cracks is not unique, the full characterization of the crack characteristics is ensured by establishing a data set, extracting crack features and applying a kernel density function. The effectiveness of data acquisition is improved, and the crack features are classified by a clustering algorithm to determine the crack grade, thereby improving the automation level and accuracy of asphalt pavement crack detection.

[0047] On the other hand, the present application also provides an asphalt pavement crack recognition system based on image features, which is used to apply the above-mentioned asphalt pavement crack recognition method based on image features, including:

[0048] A collection unit is configured to collect characteristic information of the road surface to be monitored, wherein the characteristic information includes road surface size, color difference between the road surface and the surrounding area, and road surface flatness, and determine the size information of a single image during image collection according to the characteristic information;

[0049] a judgment unit configured to divide the road surface to be monitored into regions according to the size information, collect initial image data for each region, obtain the number of cracks in the initial image data based on image recognition, and judge whether to expand the image data volume according to the number of cracks;

[0050] A processing unit is configured to, when the number of cracks is not unique, establish a data set and determine the amount of image data in the data set according to the number of cracks; when it is determined that the data set is established, extract the crack features of each image data, determine a representational representative graph based on a kernel density function, and determine regional crack features according to the representational representative graph;

[0051] The identification unit is configured to integrate all the regional crack features, classify the crack features of the road surface to be monitored by a clustering algorithm, and determine the crack grade of the road surface to be monitored.

[0052] It is understandable that the above-mentioned asphalt pavement crack identification method and system based on image features have the same beneficial effects, which will not be described in detail here. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] 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:

[0054] Figure 1 A flowchart of an asphalt pavement crack identification method based on image features provided by an embodiment of the present invention;

[0055] Figure 2 A functional block diagram of an asphalt pavement crack identification system based on image features provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0056] 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.

[0057] In some embodiments of the present application, see Figure 1As shown, a method for identifying cracks in asphalt pavement based on image features comprises:

[0058] S100: Collect characteristic information of the road surface to be monitored, the characteristic information including road surface size, color difference between the road surface and the surrounding area, and road surface flatness, and determine the size information of a single image during image acquisition based on the characteristic information.

[0059] S200: Divide the road surface to be monitored into regions according to the size information, collect initial image data for each region, obtain the number of cracks in the initial image data based on image recognition, and determine whether to expand the image data volume according to the number of cracks.

[0060] S300: When the number of cracks is not unique, a data set is established and the amount of image data in the data set is determined according to the number of cracks. When it is determined that a data set is established, crack features of each image data are extracted, a representative representation graph is determined based on a kernel density function, and regional crack features are determined according to the representative representation graph.

[0061] S400: Integrate crack characteristics of all regions, classify crack characteristics of the road surface to be monitored by clustering algorithm, and determine the crack grade of the road surface to be monitored.

[0062] Specifically, by collecting characteristic information of the road surface to be monitored, including road surface size, color difference with the surrounding area, and road surface flatness. The characteristic information provides the basis for image acquisition. According to the collected characteristic information, the size of a single image is determined. It is ensured that the captured image can fully cover the road surface features, which is convenient for subsequent analysis. According to the determined size information, the road surface to be monitored is divided into regions, and initial image data is collected for each region. Through image recognition technology, the number of cracks in each image is obtained, and it is determined whether the image data needs to be expanded according to the number of cracks to ensure that the amount of collected data matches the complexity. In the case where the number of cracks is not unique, a corresponding data set is established, and the amount of image data is determined according to the number of cracks. The crack features are extracted for each image, and the representative image is determined by the kernel density function, so as to clarify the regional crack features. Finally, the crack features of all regions are integrated, and the crack features are classified by clustering algorithm to accurately determine the crack grade of the road surface.

[0063] It is understandable that by collecting and analyzing road surface feature information and combining image processing technology, efficient and accurate identification of asphalt pavement cracks can be achieved. Compared with traditional methods, the subjectivity and error of manual detection are reduced, and the comprehensiveness and representativeness of data are improved by dynamically adjusting image acquisition strategies and data expansion. At the same time, the application of clustering algorithms ensures the accuracy of crack feature classification, making pavement maintenance decisions more accurate. The automation level and accuracy of crack detection are improved.

[0064] In some embodiments of the present application, when determining the size information of a single image during image acquisition according to the characteristic information, it includes obtaining a characteristic index of the road surface to be monitored according to the characteristic information, and determining the size information of the single image according to the characteristic index, wherein the characteristic index is calculated by the following formula:

[0065]

[0066] Among them, T represents the characteristic index, L represents the actual length of the road surface to be monitored, D1 represents the color difference measurement between the road surface and the surrounding area, and D2 represents the flatness measurement of the road surface.

[0067] In some embodiments of the present application, when determining the size information of a single image based on the feature index, it includes: comparing the feature index with a preset first preset feature index and a second preset feature index respectively, determining the size area of ​​the single image based on the comparison result, and the first preset feature index is smaller than the second preset feature index.

[0068] When the characteristic index is less than or equal to the first preset characteristic index, the size area is determined to be the first preset area. When the characteristic index is greater than the first preset characteristic index and less than or equal to the second preset characteristic index, the size area is determined to be the second preset area. When the characteristic index is greater than the second preset characteristic index, the size area is determined to be the third preset area. The first preset area is smaller than the second preset area, and the second preset area is smaller than the third preset area.

[0069] It can be understood that by combining the calculation of the characteristic index with the comparison of the preset value, the dynamic adjustment of the image acquisition size is achieved, which enhances the adaptability and flexibility of the scheme. By measuring the characteristic index, the image acquisition area can be effectively optimized for different road conditions, thereby ensuring that the collected images can more accurately reflect the road conditions. The quality of image data is improved, providing a more reliable basis for subsequent crack identification and evaluation.

[0070] In some embodiments of the present application, when determining whether to expand the image data volume according to the number of cracks, it includes: when the number of cracks in the initial image data is unique, it is determined not to expand the image data, and the crack features of all regions are integrated, and the crack features of the road surface to be monitored are classified by a clustering algorithm to determine the crack level of the road surface to be monitored. When the number of cracks in the initial image data is not unique, it is determined to expand the image data volume.

[0071] In some embodiments of the present application, when determining to expand the image data volume, establishing a data set and determining the image data volume in the data set according to the number of cracks, it includes:

[0072]

[0073] Wherein, N represents the amount of image data, N is rounded up, C represents the constant factor, R represents the number of cracks, Q represents the light intensity for collecting the initial image data, and Q0 represents the average light intensity.

[0074] In some embodiments of the present application, when determining the amount of image data and establishing a data set, it also includes: collecting image data at different angles for the same area during the same period of time.

[0075] It can be understood that the judgment is made based on the uniqueness of the number of cracks in the initial image data. When the number of cracks is unique, it is determined not to expand the image data, and the regional crack features are directly extracted and classified by the clustering algorithm. This effectively reduces unnecessary data redundancy and improves processing efficiency. When the number of cracks is not unique, the new image data volume is calculated. It ensures that sufficient information can still be obtained to support subsequent analysis in a complex environment. Image data is collected from different angles in the same area and at the same time. Providing more diverse perspectives helps to capture multi-dimensional information of crack features and further improve the accuracy of recognition. By judging the uniqueness of the number of cracks and dynamically adjusting the image data volume, the adequacy and efficiency of data collection are effectively balanced. When the number of cracks is not unique, by reasonably expanding the data set, it is ensured that sufficient samples can still be obtained under complex road conditions, thereby enhancing the accuracy of subsequent crack recognition. In addition, the strategy of multi-angle acquisition further enriches the data features and improves the comprehensive understanding of crack features.

[0076] In some embodiments of the present application, determining a characterization representative graph based on a kernel density function and determining regional crack characteristics according to the characterization representative graph include:

[0077] Perform kernel density estimation on the data set and determine the image data with the highest probability as the representative image.

[0078] When performing kernel density estimation, the Gaussian kernel density function is:

[0079]

[0080] Among them, n represents the total number of image data in the data set, h represents the smoothing bandwidth, and s i represents the crack characteristics of the i-th image data, and x represents any variable in the overall data.

[0081] In some embodiments of the present application, the smoothing bandwidth is calculated by the following formula:

[0082]

[0083] Where h represents the smoothing bandwidth, σ represents the standard deviation of the crack characteristics of all image data in the data set, β1 represents the data skewness, β2 represents the data kurtosis, n represents the total number of image data in the data set, and k represents the adjustment coefficient.

[0084] The data skewness is calculated by the following formula:

[0085]

[0086] The kurtosis of the data is calculated by the following formula:

[0087]

[0088] Among them, n represents the total number of image data in the dataset, s i represents the crack characteristics of the i-th image data, represents the mean of the crack characteristics of all image data, and σ represents the standard deviation of the crack characteristics of all image data.

[0089] It can be understood that kernel density estimation is performed on the image data set to identify the probability distribution of crack features. In this process, the Gaussian kernel density function is used to evaluate the density of crack features in each image and generate a smooth probability distribution map. Through kernel density estimation, the distribution of crack features is effectively converted into a probability model, which enhances the ability to analyze data. Using the Gaussian kernel density function, the most representative crack features in the data set can be accurately identified, providing a refined analysis basis. The dynamic calculation of the smoothing bandwidth ensures the accuracy of the estimation and adapts to the needs of different data characteristics. The accuracy of crack feature identification is further improved.

[0090] In some embodiments of the present application, the crack characteristics of the road surface to be monitored are classified by a clustering algorithm to determine the crack grade of the road surface to be monitored, including:

[0091] S1: Initialize K centroids in the classified historical pavement image data, assign the crack features of the road surface to be monitored to the nearest centroid, and form K clusters.

[0092] S2: Recalculate the centroid of each cluster.

[0093] S3: Repeat S1 and S2 until the center of mass no longer changes or the preset number of repetitions is completed.

[0094] S4: The classification and crack grade corresponding to the historical road surface image data in the cluster where the representative graph is located are used as the evaluation result of the crack characteristics of the road surface to be monitored.

[0095] The centroid calculation formula is as follows:

[0096]

[0097] Among them, Zk represents the centroid of the kth cluster, |Ck| represents the number of data in the kth cluster, and Ti represents the feature vector of the i-th data.

[0098] It is understandable that by using the K-means clustering algorithm, the crack characteristics are automatically classified and evaluated, which improves the efficiency and accuracy of crack grade identification. Using the classification information in the historical image data, the new data can be accurately classified with the help of existing data, reducing the need for manual intervention and improving the consistency of the evaluation. The data-driven clustering method optimizes the decision-making process.

[0099] In the above embodiments, the dimensional information during image acquisition is determined by comprehensively considering the road surface feature information (such as road surface size, color difference and flatness), and the road surface to be monitored is finely divided into regions, thereby ensuring comprehensive image acquisition of cracks in each region. Based on image recognition technology, the number of cracks can be accurately obtained, and it can be flexibly determined whether the amount of image data needs to be expanded to cope with complex road conditions. When the number of cracks is not unique, the full characterization of the crack characteristics is ensured by establishing a data set, extracting crack features and applying a kernel density function. The effectiveness of data acquisition is improved, and the crack features are classified and the crack grade is determined through a clustering algorithm, thereby improving the automation level and accuracy of asphalt pavement crack detection.

[0100] In another preferred embodiment based on the above embodiment, refer to Figure 2 As shown, this embodiment provides an asphalt pavement crack recognition system based on image features, which is used to apply the above-mentioned asphalt pavement crack recognition method based on image features, including:

[0101] A collection unit is configured to collect characteristic information of the road surface to be monitored, the characteristic information including road surface size, color difference between the road surface and the surrounding area, and road surface flatness, and determine the size information of a single image during image collection according to the characteristic information;

[0102] A judgment unit is configured to divide the road surface to be monitored into regions according to the size information, collect initial image data for each region, obtain the number of cracks in the initial image data based on image recognition, and judge whether to expand the image data volume according to the number of cracks;

[0103] The processing unit is configured to establish a data set and determine the amount of image data in the data set according to the number of cracks when the number of cracks is not unique; when it is determined that the data set is established, extract the crack characteristics of each image data, determine the representation representative map based on the kernel density function, and determine the regional crack characteristics according to the representation representative map;

[0104] The identification unit is configured to integrate the crack characteristics of all regions, classify the crack characteristics of the road surface to be monitored through a clustering algorithm, and determine the crack grade of the road surface to be monitored.

[0105] It is understandable that by comprehensively considering the road surface feature information (such as road surface size, color difference and flatness), determining the size information during image acquisition, and finely dividing the monitored road surface into regions, comprehensive image acquisition of cracks in each region is ensured. Based on image recognition technology, the number of cracks can be accurately obtained, and it can be flexibly judged whether the image data volume needs to be expanded to cope with complex road conditions. When the number of cracks is not unique, the full characterization of the crack characteristics is ensured by establishing a data set, extracting crack features and applying a kernel density function. The effectiveness of data acquisition is improved, and the crack features are classified and the crack grade is determined through a clustering algorithm, which improves the automation level and accuracy of asphalt pavement crack detection.

[0106] 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 adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt 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.

[0107] The present application is described with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to 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 generate 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.

[0108] 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 comprising 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.

[0109] 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.

[0110] 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 method for identifying cracks in asphalt pavement based on image features, characterized in that: include: Collect characteristic information of the road surface to be monitored, the characteristic information including road surface size, color difference between the road surface and the surrounding area, and road surface flatness, and determine the size information of a single image during image acquisition based on the characteristic information; Divide the road surface to be monitored into regions according to the size information, collect initial image data for each region, obtain the number of cracks in the initial image data based on image recognition, and determine whether to expand the image data volume according to the number of cracks; When the number of cracks is not unique, establishing a data set and determining the amount of image data in the data set according to the number of cracks; When it is determined that the data set is to be established, crack features of each image data are extracted, a characterization representative graph is determined based on a kernel density function, and regional crack features are determined according to the characterization representative graph; All the regional crack characteristics are integrated, the crack characteristics of the road surface to be monitored are classified by a clustering algorithm, and the crack grade of the road surface to be monitored is determined.

2. The asphalt pavement crack identification method based on image features according to claim 1 is characterized in that: The determining of the size information of a single image during image acquisition according to the characteristic information includes: obtaining a characteristic index of the road surface to be monitored according to the characteristic information, and determining the size information of the single image according to the characteristic index, wherein the characteristic index is calculated by the following formula: Among them, T represents the characteristic index, L represents the actual length of the road surface to be monitored, D1 represents the color difference measurement between the road surface and the surrounding area, and D2 represents the flatness measurement of the road surface.

3. The asphalt pavement crack identification method based on image features according to claim 2 is characterized in that: Determining the size information of the single image according to the characteristic index includes: The characteristic index is compared with a first preset characteristic index and a second preset characteristic index respectively, and the size area of ​​the single image is determined according to the comparison result, wherein the first preset characteristic index is smaller than the second preset characteristic index; When the characteristic index is less than or equal to the first preset characteristic index, the size area is determined to be the first preset area; when the characteristic index is greater than the first preset characteristic index and less than or equal to the second preset characteristic index, the size area is determined to be the second preset area; when the characteristic index is greater than the second preset characteristic index, the size area is determined to be the third preset area; the first preset area is smaller than the second preset area, and the second preset area is smaller than the third preset area.

4. The asphalt pavement crack identification method based on image features according to claim 1 is characterized in that: When judging whether to expand the image data volume according to the number of cracks, it includes: When the number of cracks in the initial image data is unique, it is determined not to expand the image data, and all the regional crack features are integrated, and the crack features of the road surface to be monitored are classified by a clustering algorithm to determine the crack grade of the road surface to be monitored; When the number of cracks in the initial image data is not unique, the amount of expanded image data is determined.

5. The asphalt pavement crack identification method based on image features according to claim 4 is characterized in that: When determining to expand the image data volume, establishing a data set and determining the image data volume in the data set according to the number of cracks, it includes: Wherein, N represents the amount of image data, N is rounded up, C represents the constant factor, R represents the number of cracks, Q represents the light intensity for collecting the initial image data, and Q0 represents the average light intensity.

6. The asphalt pavement crack identification method based on image features according to claim 4 is characterized in that: When determining the image data amount, establishing the data set also includes: Image data is collected from different angles in the same area during the same period of time.

7. The asphalt pavement crack identification method based on image features according to claim 1 is characterized in that: Determining a representative image based on a kernel density function, and determining regional crack characteristics based on the representative image, includes: Performing kernel density estimation on the data set to determine the image data with the highest probability as the representation representative image; When performing the kernel density estimation, the Gaussian kernel density function is: Among them, n represents the total number of image data in the data set, h represents the smoothing bandwidth, and s i represents the crack characteristics of the i-th image data, and x represents any variable in the overall data.

8. The asphalt pavement crack identification method based on image features according to claim 7 is characterized in that: The smoothing bandwidth is calculated by the following formula: Where h represents the smoothing bandwidth, σ represents the standard deviation of the crack characteristics of all image data in the data set, β1 represents the data skewness, β2 represents the data kurtosis, n represents the total number of image data in the data set, and k represents the adjustment coefficient; The data skewness is calculated by the following formula: The data kurtosis is calculated by the following formula: Among them, n represents the total number of image data in the dataset, s i represents the crack characteristics of the i-th image data, represents the mean of the crack characteristics of all image data, and σ represents the standard deviation of the crack characteristics of all image data.

9. The asphalt pavement crack identification method based on image features according to claim 1 is characterized in that: Classifying the crack characteristics of the road surface to be monitored by a clustering algorithm to determine the crack grade of the road surface to be monitored includes: S1: Initialize K centroids in the classified historical road surface image data, and assign the crack features of the road surface to be monitored to the nearest centroid to form K clusters; S2: Recalculate the centroid of each cluster; S3: Repeat S1 and S2 until the center of mass no longer changes or the preset number of repetitions is completed; S4: taking the classification and crack grade corresponding to the historical road surface image data in the cluster where the characterizing representative graph is located as the evaluation result of the crack characteristics of the road surface to be monitored; The centroid calculation formula is as follows: Among them, Zk represents the centroid of the kth cluster, |Ck| represents the number of data in the kth cluster, and Ti represents the feature vector of the i-th data.

10. An asphalt pavement crack recognition system based on image features, used for applying the asphalt pavement crack recognition method based on image features as claimed in any one of claims 1 to 9, characterized in that: include: A collection unit is configured to collect characteristic information of the road surface to be monitored, wherein the characteristic information includes road surface size, color difference between the road surface and the surrounding area, and road surface flatness, and determine the size information of a single image during image collection according to the characteristic information; a judgment unit configured to divide the road surface to be monitored into regions according to the size information, collect initial image data for each region, obtain the number of cracks in the initial image data based on image recognition, and judge whether to expand the image data volume according to the number of cracks; a processing unit configured to, when the number of cracks is not unique, establish a data set and determine the amount of image data in the data set according to the number of cracks; When it is determined that the data set is to be established, crack features of each image data are extracted, a characterization representative graph is determined based on a kernel density function, and regional crack features are determined according to the characterization representative graph; The identification unit is configured to integrate all the regional crack features, classify the crack features of the road surface to be monitored by a clustering algorithm, and determine the crack grade of the road surface to be monitored.

Images