Method and system for detecting content of glass beads in pavement marking

Abstract

The invention discloses a method and a system for detecting the content of glass beads in pavement marking, and belongs to the field of measurement data analysis, the detection method comprises the following steps: scanning glass beads with different refractive indexes by using a near infrared spectrometer, and extracting spectral data of the glass beads with different refractive indexes to construct a comparison database; the method comprises the following steps: scanning a to-be-detected sample by using a near-infrared spectrometer to obtain spectral data, constructing a spectral database, preprocessing the spectral data obtained by scanning, and performing feature extraction on the preprocessed spectral data; and analyzing the spectral peak signal characteristic data and the data in the spectral database by using a Lasso regression algorithm to obtain the final content of the glass beads. The whole detection process is convenient and rapid, and more accurate detection can be realized in combination with a data algorithm.

Description

Technical Field

[0001] The invention relates to the field of measurement data analysis, and in particular to a method and system for detecting glass bead content in road markings. Background Art

[0002] Road marking is one of the important traffic safety facilities. Clear and standardized road markings can provide clear and safe guidance to traffic participants. In road markings, there are certain standards and requirements for the content of glass beads to ensure its reflective performance and durability. Different countries and regions may have different standards, but in general, the content of glass beads should account for 10%-30% of the total weight of the road marking paint. This ratio can be adjusted according to different traffic conditions and paint types. The distribution density of glass beads, glass beads should be evenly distributed on the surface of the road marking, and there is usually a minimum number of glass bead particles per square meter of marking. For example, some standards require that the density of glass beads per square meter of marking should not be less than 500. At the same time, due to the need for night indication, road markings are also required to have good reflective performance, and the amount of glass beads added needs to meet certain reflective performance standards, such as the reflection coefficient (usually expressed as Luminance Intensity) at a certain distance and angle. These data are closely related to the content and distribution density of glass beads in road markings. Before road coating construction, it is particularly important to detect the glass bead content in the coating and measure its various properties for the final marking effect.

[0003] At present, the domestic detection and analysis methods for the glass bead content of road markings are mainly achieved through chemical composition analysis. For example, the invention titled "A Method for Detecting the Glass Bead Content in Hot-melt Reflective Road Markings" published on February 3, 2023, and the patent document with publication number CN115683918A records a method for detecting the glass bead content in hot-melt reflective road markings. This method detects the glass bead content in hot-melt reflective road markings by chemical analysis, prepares the required raw materials, equipment and utensils in advance, and then calculates the glass bead content through chemical analysis and formula calculation. Although this method reduces the use of organic solvents and improves environmental friendliness. However, it still requires cumbersome operation processes, long waiting times, and complex calculations to obtain the glass bead content. And it is impossible to calculate the distribution density of glass beads and the minimum number of glass bead particles per square meter of markings and other related information. Summary of the invention

[0004] One of the purposes of the present invention is to provide a method for detecting the glass bead content in road markings and to solve the shortcomings of the prior art, such as slow detection speed and little detection information obtained.

[0005] The present invention is achieved through the following technical scheme. A method for detecting the content of glass beads in road markings comprises the following steps: S100, using a near-infrared spectrometer to scan glass beads with different refractive indices, extracting spectral data of the glass beads with different refractive indices to construct a comparison database; S200, using a near-infrared spectrometer to scan a sample to be detected, obtaining spectral data, and constructing a spectral database, wherein the sample to be detected is obtained by uniformly coating 10 to 20 g of prepared road marking paint on a square substrate with a side length of 10 cm; S300, preprocessing the spectral data obtained by scanning, and extracting features from the preprocessed spectral data; S400, using a Lasso regression algorithm to analyze the characteristic data of the spectral peak signal and the data in the spectral database to obtain the final glass bead content.

[0006] Furthermore, the detection method also includes, S500, an inversion step, constructing a glass bead content prediction inversion model based on the data in the comparison database and the spectral database.

[0007] Furthermore, glass beads with different refractive indices include low refractive index glass beads, medium refractive index glass beads and high refractive index glass beads, among which glass beads with a refractive index of 1.50≤RI<1.70 are low refractive index glass beads; glass beads with a refractive index of 1.70≤RI<1.90 are medium refractive index glass beads; and glass beads with a refractive index of RI>1.90 are high refractive index glass beads, where RI is the refractive index.

[0008] Furthermore, the preprocessing includes using a smoothing filter to reduce noise on the scanned spectral data, and then standardizing the spectral data by Box-Cox transformation.

[0009] Furthermore, the feature extraction includes the following sub-steps: S310, dividing the spectral data into two-dimensional grids, dividing them into 100 two-dimensional grids with a side length of 1 cm, and identifying each square in the grid to obtain the identification ID of each cube; S320, using a peak detection algorithm to identify the characteristic peaks in each two-dimensional grid, and matching the corresponding characteristic peaks with the identification ID; S330, extracting the spectral peak signals of glass beads with different refractive indices from the comparison database, and comparing the characteristic peaks one by one according to the identification ID, and extracting the qualified characteristic peaks, and calculating the spectral peak signal characteristic data based on the extracted characteristic peak data.

[0010] Furthermore, the peak signal characteristic data includes peak area, peak height and peak width; wherein the peak area is calculated by the following formula: , where A is the area of ​​the peak; f (x) is the relationship between the peak intensity and the wavelength; x1 is the starting position of the peak, and x2 is the ending position of the peak; y i is the spectral intensity of the scanning light; x iis the wavelength of the scanning light; n is the number of spectral peak signals that meet the characteristics; the spectral peak height is calculated by the following formula, H=y max −y baseline , where H is the height of the peak; y max is the intensity of the highest point of the spectrum peak; baseline is the baseline value of the peak; the peak width is described by calculating the peak half-height width (FWHM), that is, at half the peak height, the width of the abscissa is, FWHM=x right −x left , where x right and x left is the wavelength position corresponding to half the peak height.

[0011] Furthermore, the inversion step may also include the following sub-steps: S510, determining the target data range for inversion based on the data in the spectral database; S520, then using the spectral peak signal characteristic data to construct a spectral peak signal data dictionary; S530, according to the principle of matching pursuit method, matching the data to be inverted with the characteristic data of the spectral peak signal data dictionary point by point, extracting the data pairs with the greatest matching degree to form an inverted spectral peak signal data body; S540, constructing a deep learning network for inverting the actual road surface marking reflection coefficient based on the spectral peak signal data dictionary, the inverted spectral peak signal data body and the comparison database, and training the deep learning network using the data sets in the comparison database and the spectral database to obtain a glass bead content prediction inversion model; S550, inputting the actual road surface marking area into the inversion model to obtain the reflection coefficient of the road surface marking after the final construction.

[0012] Another aspect of the present invention provides a detection system for glass bead content in road markings. The detection system includes a comparison database construction unit, a spectrum database construction unit, a data preprocessing unit, a data analysis unit and an inversion unit.

[0013] The comparison database construction unit is configured to use a near-infrared spectrometer to scan glass beads with different refractive indices, extract spectral data of glass beads with different refractive indices to construct a comparison database, wherein glass beads with a refractive index of 1.50≤RI<1.70 are low-refractive-index glass beads; glass beads with a refractive index of 1.70≤RI<1.90 are medium-refractive-index glass beads; and glass beads with a refractive index of RI>1.90 are high-refractive-index glass beads, where RI is the refractive index. The spectral database construction unit is configured to evenly coat 10-20g of the prepared road marking paint on a square substrate with a side length of 10cm to obtain a sample to be tested, scan the sample to be tested with a near-infrared spectrometer, obtain spectral data, and construct a spectral database. The data preprocessing unit is connected to the comparison database construction unit and the spectrum database construction unit respectively, and is configured to preprocess the spectral data obtained by scanning, the preprocessing includes using smoothing filtering to reduce noise on the spectral data obtained by scanning, and then standardizing the spectral data through Box-Cox transformation, and extracting features from the preprocessed spectral data, the feature extraction includes dividing the spectral data into two-dimensional grids, dividing them into 100 two-dimensional grids with a side length of 1 cm, and marking each square in the grid to obtain the identification ID of each cube; using the peak detection algorithm to identify the characteristic peak in each two-dimensional grid, and matching the corresponding characteristic peak with the identification ID; extracting the spectrum peak signal of glass beads with different refractive indices from the comparison database, and comparing the characteristic peaks one by one according to the identification ID, identifying the characteristic peaks with a spectrum peak signal similarity of more than 90% in the comparison database as qualified characteristic peaks and extracting them, and calculating the spectrum peak signal characteristic data according to the extracted characteristic peak data. The data analysis unit is connected to the comparison database construction unit, and is configured to use the Lasso regression algorithm to analyze the spectrum peak signal characteristic data and the data in the spectrum database to obtain the final glass bead content. The inversion unit is connected to the data preprocessing unit and the data analysis unit respectively, and is configured to determine the target data range of inversion according to the data in the spectral database; then use the spectral peak signal characteristic data to construct a spectral peak signal data dictionary; according to the principle of matching pursuit method, the data to be inverted are matched point by point with the characteristic data of the spectral peak signal data dictionary, and the data pairs with the largest matching degree are extracted to form an inverted spectral peak signal data body; based on the spectral peak signal data dictionary, the inverted spectral peak signal data body and the comparison database, a deep learning network for inverting the reflection coefficient of the actual road surface marking is constructed, and the deep learning network is trained using the above data set to obtain a glass bead content prediction inversion model; the actual road surface marking area is input into the inversion model to obtain the reflection coefficient of the road surface marking after the final construction.

[0014] In another aspect, the present invention provides a computer device, comprising: a processor; a memory storing a computer program, and when the computer program is executed by the processor, the method for detecting the glass bead content in the road marking as described above is implemented.

[0015] In another aspect, the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the method for detecting the glass bead content in a road marking as described above.

[0016] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0017] 1. The present invention changes the original chemical detection method into a physical optical detection method by optical scanning, which does not require cumbersome chemical operation processes and various chemical reagents. The entire detection process is convenient and fast, and combined with data algorithms, more accurate detection can be achieved. Compared with chemical detection methods, the detection accuracy is greatly improved.

[0018] 2. The present invention collects spectral data and analyzes the data to detect the distribution density of glass beads in the paint, and by constructing various data into an inversion model, it can predict the reflection coefficient of the road marking after actual construction, greatly improving the construction efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings:

[0020] Figure 1 This is a flow chart of the method provided in Example 1 of the present invention.

[0021] Figure 2 A schematic diagram of the system structure provided in Example 2 of the present invention.

[0022] Figure 3 A schematic diagram of the computer device structure provided in Example 3 of the present invention.

[0023] Description of reference numerals:

[0024] 100 - a detection system for glass bead content in road markings, 110 - a comparison database construction unit, 120 - a spectrum database construction unit, 130 - a data preprocessing unit, 140 - a data analysis unit, 150 - an inversion unit, 200 - a computer device, 210 - a memory, 220 - a processor. DETAILED DESCRIPTION

[0025] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0026] Embodiment 1:

[0027] Step 1: According to the provisions of the national standard for glass beads for road marking, glass beads are divided into low refractive index glass beads, medium refractive index glass beads and high refractive index glass beads according to their different refractive indices. The refractive index (RI) of low refractive index glass beads is 1.50≤RI<1.70; the refractive index (RI) of medium refractive index glass beads is 1.70≤RI<1.90; the refractive index (RI) of high refractive index glass beads is RI>1.90. Use a near-infrared spectrometer to scan three glass beads with different refractive indices in advance, extract the spectral data of glass beads with different refractive indices to build a comparison database.

[0028] Step 2: Take 10-20g of the prepared road marking paint and evenly apply it on a square substrate with a side length of 10cm to obtain a sample to be tested. Use a near-infrared spectrometer to scan the sample to be tested, obtain spectral data, and build a spectral database.

[0029] Specifically, the square substrate may be an opaque acrylic plastic plate or a marble plate.

[0030] Step 3: preprocess the spectral data obtained by scanning, and then perform feature extraction on the preprocessed spectral data.

[0031] Specifically, the preprocessing includes first using smoothing filtering to reduce the noise of the scanned spectral data, and then standardizing the spectral data through Box-Cox transformation.

[0032] It should be noted that the reason why the Box-Cox transformation is chosen to standardize the spectral data is that there may be large variance fluctuations in the spectral data. The Box-Cox transformation can stabilize the variance of the data, making the variability of the data more consistent at different wavelengths, thereby improving the reliability of the analysis. In addition, the road marking material is evenly coated on the substrate, and the distribution of the glass beads evenly coated on the plane is also evenly distributed. However, since the paint and the glass beads have different reflection frequencies, the high reflectivity of the glass beads will inevitably lead to a large number of spectral peak signals in the spectral data, so that the spectral data obtained after the spectral scanning may show a skewed distribution. The Box-Cox transformation can reduce or eliminate this skewness, making the data distribution more symmetrical, which helps to improve the accuracy of the analysis results. In the subsequent steps, the spectral data needs to be used for inversion modeling, and the Box-Cox transformation can convert the data into a form that is more suitable for modeling, thereby improving the performance of the model, so that the model can better capture the patterns and trends in the data. And because the Box-Cox transformation is reversible, it can provide effective inversion data for the next step of inversion and provide good support for the construction of the inversion model.

[0033] Specifically, feature extraction includes dividing the spectral data into two-dimensional grids, dividing them into 100 two-dimensional grids with a side length of 1 cm, and marking each square in the grid to obtain the identification ID of each cube.

[0034] A peak detection algorithm is used to identify the characteristic peaks in each two-dimensional grid, and the corresponding characteristic peaks are matched with identification IDs.

[0035] The spectrum peak signals of glass beads with different refractive indices are extracted from the comparison database, and the characteristic peaks are compared one by one according to the identification ID, and the qualified characteristic peaks are extracted, and the spectrum peak signal characteristic data is obtained according to the extracted qualified characteristic peak data. Specifically, the qualified characteristic peak is a characteristic peak with a similarity of more than 90% with the spectrum peak signal of the comparison database.

[0036] Specifically, the spectrum peak signal characteristic data includes spectrum peak area, spectrum peak height and spectrum peak width.

[0037] The peak area is calculated by the following formula:

[0038] ,

[0039] Where A is the area of ​​the peak; f (x) is the relationship between the peak intensity and the wavelength; x1 is the starting position of the peak, x2 is the ending position of the peak; y i is the spectral intensity of the scanning light; x i is the wavelength of the scanning light; n is the number of spectral peak signals that meet the characteristics.

[0040] The peak height is calculated by the following formula:

[0041] H=y max −y baseline , where H is the height of the peak; y max is the intensity of the highest point of the spectrum peak; baseline is the baseline value of the peak.

[0042] The peak width is described by calculating the peak half-height width (FWHM), that is, the width of the abscissa at half the peak height is,

[0043] FWHM=x right −x left , where x right and x left is the wavelength position corresponding to half the peak height.

[0044] Step 4: Use the Lasso regression algorithm to analyze the peak signal characteristic data to obtain the final glass bead content.

[0045] It should be noted that the reason why the Lasso regression algorithm is used to analyze the characteristic data of the spectral peak signal is that Lasso regression will generate a relatively simple model, which is easier to implement and maintain in practical applications, and has higher computational efficiency than Lasso regression. In addition, Lasso regression has the ability of automatic feature selection. It can automatically remove features that are not important for prediction by applying L1 regularization to the data feature weights, making the coefficients of some features zero. This is especially useful for redundant and noisy features that may exist in spectral data. In addition, Lasso regression will select fewer features, making it easier to understand which spectral features play a key role in predicting the glass bead content. This facilitates the construction of subsequent inversion models.

[0046] The purpose of detecting the glass bead content in the road marking can be achieved through the above steps. Based on the data obtained in the above steps, the present invention also provides a method for constructing an inversion model through various data obtained in the above steps. By using the inversion model, the reflection coefficient of the road marking after construction can be inferred based on the detected glass bead content.

[0047] The details are as described in the following steps.

[0048] Step 5: Based on the data in the comparison database and the spectral database, a glass bead content prediction inversion model is constructed.

[0049] Specifically, firstly, the target data range for inversion is determined based on the data in the spectral database;

[0050] Then, the spectrum peak signal data dictionary is constructed using the spectrum peak signal characteristic data;

[0051] According to the principle of matching pursuit method, the data to be inverted are matched point by point with the characteristic data of the spectrum peak signal data dictionary, and the data pairs with the greatest matching degree are extracted to form the inverted spectrum peak signal data body;

[0052] Based on the spectral peak signal data dictionary, the inverted spectral peak signal data body and the comparison database, a deep learning network for inverting the actual road marking reflection coefficient is constructed, and the deep learning network is trained using the data sets in the comparison database and the spectral database to obtain a glass bead content inversion model;

[0053] The actual road marking area is input into the inversion model to obtain the reflection coefficient of the road marking after final construction.

[0054] It should be noted that, unlike the usual chemical detection, optical scanning is used to detect the glass bead content in the road markings. It does not require a large amount of chemical reagents and cumbersome chemical operation steps. After the entire model system is built, detailed glass bead content data can be obtained by simply applying the prepared paint on the substrate and scanning it. As the detection proficiency improves, the constructed database can also make the final prediction results more accurate. More conveniently, the inversion model is constructed through various data, and the reflection coefficient of the road marking after actual construction can be predicted based on the test results and the specific road marking construction area. Construction personnel can pre-adjust the glass bead content in the paint according to the inversion results, which can facilitate the operation of construction personnel and greatly improve construction efficiency.

[0055] Embodiment 2,

[0056] like Figure 2 As shown, the detection system 100 for glass bead content in road markings includes a comparison database construction unit 110 , a spectrum database construction unit 120 , a data preprocessing unit 130 , a data analysis unit 140 and an inversion unit 150 .

[0057] The comparison database construction unit 110 is configured to use a near-infrared spectrometer to scan glass beads with different refractive indices, extract spectral data of glass beads with different refractive indices to construct a comparison database, wherein glass beads with a refractive index of 1.50≤RI<1.70 are low refractive index glass beads; glass beads with a refractive index of 1.70≤RI<1.90 are medium refractive index glass beads; and glass beads with a refractive index of RI>1.90 are high refractive index glass beads, where RI is the refractive index.

[0058] The spectral database construction unit 120 is configured to uniformly apply 10-20 g of the prepared road marking paint on a square substrate with a side length of 10 cm to obtain a sample to be tested, use a near-infrared spectrometer to scan the sample to be tested, obtain spectral data, and construct a spectral database.

[0059] The data preprocessing unit 130 is connected to the comparison database construction unit 110 and the spectrum database construction unit 120 respectively, and is configured to preprocess the spectral data obtained by scanning. The preprocessing includes using smoothing filtering to reduce the noise of the scanned spectral data, and then standardizing the spectral data through Box-Cox transformation, and extracting features from the preprocessed spectral data.

[0060] Specifically, feature extraction includes dividing the spectral data into two-dimensional grids, dividing them into 100 two-dimensional grids with a side length of 1 cm, and identifying each square in the grid to obtain the identification ID of each cube; using a peak detection algorithm to identify the characteristic peaks in each two-dimensional grid, and matching the corresponding characteristic peaks with the identification ID; extracting the spectral peak signals of glass beads with different refractive indices from a comparison database, and comparing the characteristic peaks one by one according to the identification ID, identifying the characteristic peaks with a spectral peak signal similarity of more than 90% in the comparison database as qualified characteristic peaks and extracting them, and calculating the spectral peak signal characteristic data based on the extracted characteristic peak data;

[0061] The data analysis unit 140 is connected to the data preprocessing unit 130 and the spectrum database construction unit 120 respectively, and is configured to use the Lasso regression algorithm to analyze the spectrum peak signal characteristic data and the data in the spectrum database to obtain the final glass bead content.

[0062] The inversion unit 150 is connected to the data preprocessing unit 130 and the data analysis unit 140 respectively, and is configured to determine the target data range for inversion according to the data in the spectral database; then use the spectral peak signal characteristic data to construct a spectral peak signal data dictionary; according to the principle of matching pursuit method, the data to be inverted is matched point by point with the characteristic data of the spectral peak signal data dictionary, and the data pairs with the greatest matching degree are extracted to form an inverted spectral peak signal data body; based on the spectral peak signal data dictionary, the inverted spectral peak signal data body and the comparison database, a deep learning network for inverting the reflection coefficient of the actual road surface marking is constructed, and the data sets in the comparison database and the spectral database are used to train the deep learning network to obtain a glass bead content prediction inversion model; the actual road surface marking area is input into the inversion model to obtain the reflection coefficient of the road surface marking after the final construction.

[0063] Embodiment 3,

[0064] like Figure 3As shown, a computer device 200 includes a memory 210 and a processor 220. The memory 210 stores a computer program. When the computer program is executed by the processor, the method for detecting the glass bead content in the road marking as described in Example 1 is implemented.

[0065] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for detecting the content of glass beads in road markings, characterized in that: The detection method comprises: S100, using a near-infrared spectrometer to scan glass beads with different refractive indices, extracting spectral data of the glass beads with different refractive indices to build a comparison database; S200, use a near infrared spectrometer to scan the sample to be tested, obtain spectral data, and build a spectral database, The sample to be tested is obtained by evenly coating 10-20 g of the prepared road marking paint on a square substrate with a side length of 10 cm; S300, preprocessing the spectral data obtained by scanning, and extracting features from the preprocessed spectral data; S400, using Lasso regression algorithm to analyze the spectral peak signal characteristic data and the data in the spectral database to obtain the final glass bead content.

2. The method for detecting the glass bead content in road markings according to claim 1, characterized in that: The detection method further comprises: S500, inversion step, constructing a glass bead content prediction inversion model based on the data in the comparison database and the spectrum database.

3. The method for detecting the glass bead content in road markings according to claim 1, characterized in that: The glass beads with different refractive indices include low refractive index glass beads, medium refractive index glass beads and high refractive index glass beads. Among them, glass beads with a refractive index of 1.50≤RI<1.70 are low refractive index glass beads; glass beads with a refractive index of 1.70≤RI<1.90 are medium refractive index glass beads; and glass beads with a refractive index of RI>1.90 are high refractive index glass beads, where RI is the refractive index.

4. The method for detecting the glass bead content in road markings according to claim 1, characterized in that: The preprocessing includes using smoothing filtering to reduce noise on the spectral data obtained by scanning, and then standardizing the spectral data through Box-Cox transformation.

5. The method for detecting the glass bead content in road markings according to claim 1, characterized in that: The feature extraction includes the following sub-steps: S310, dividing the spectral data into two-dimensional grids, dividing them into 100 two-dimensional grids with a side length of 1 cm, and marking each square in the grid to obtain an identification ID of each cube; S320, using a peak detection algorithm to identify characteristic peaks in each two-dimensional grid, and matching the corresponding characteristic peaks with identification IDs; S330, extracting the spectrum peak signals of glass beads with different refractive indices from the comparison database, and comparing the characteristic peaks one by one according to the identification ID, and extracting the characteristic peaks that pass the comparison, and calculating the spectrum peak signal characteristic data according to the extracted characteristic peak data.

6. The method for detecting the glass bead content in road markings according to claim 5, characterized in that: The spectrum peak signal characteristic data includes spectrum peak area, spectrum peak height and spectrum peak width; The peak area is calculated by the following formula: ， Where A is the area of ​​the peak; f (x) is the relationship between the peak intensity and the wavelength; x1 is the starting position of the peak, x2 is the ending position of the peak; y i is the spectral intensity of the scanning light; x i is the wavelength of the scanning light; n is the number of spectral peak signals that meet the characteristics; The peak height is calculated by the following formula: H=y max −y baseline , where H is the height of the peak; y max is the intensity of the highest point of the spectrum peak; baseline is the baseline value of the peak; The peak width (FWHM) is the width of the abscissa at half the peak height. FWHM=x right −x left , where x right and x left is the wavelength position corresponding to half the peak height.

7. The method for detecting the glass bead content in road markings according to claim 2, characterized in that: The inversion step also includes the following sub-steps: S510, determining a target data range for inversion according to data in a spectral database; S520, then constructing a spectrum peak signal data dictionary using the spectrum peak signal characteristic data; S530, according to the principle of matching pursuit method, matching the data to be inverted with the characteristic data of the spectrum peak signal data dictionary point by point, extracting the data pair with the greatest matching degree to form an inverted spectrum peak signal data volume; S540, constructing a deep learning network for inverting the actual road marking reflection coefficient based on the spectral peak signal data dictionary, the inverted spectral peak signal data volume and the comparison database, and training the deep learning network using the data sets in the comparison database and the spectral database to obtain a glass bead content prediction inversion model; S550: Input the actual road marking area into the inversion model to obtain the reflection coefficient of the road marking after final construction.

8. A system for detecting the content of glass beads in road markings, characterized in that: The detection system comprises: Comparison database construction unit, spectrum database construction unit, data preprocessing unit, data analysis unit and inversion unit, wherein, The comparison database construction unit is configured to use a near-infrared spectrometer to scan glass beads with different refractive indices, extract spectral data of the glass beads with different refractive indices to construct a comparison database, wherein glass beads with a refractive index of 1.50≤RI<1.70 are low-refractive-index glass beads; glass beads with a refractive index of 1.70≤RI<1.90 are medium-refractive-index glass beads; and glass beads with a refractive index of RI>1.90 are high-refractive-index glass beads, where RI is the refractive index; The spectral database construction unit is configured to uniformly coat 10-20 g of the prepared road marking paint on a square substrate with a side length of 10 cm to obtain a sample to be tested, scan the sample to be tested using a near-infrared spectrometer to obtain spectral data, and construct a spectral database; The data preprocessing unit is connected to the comparison database construction unit and the spectrum database construction unit respectively, and is configured to preprocess the spectral data obtained by scanning. The preprocessing includes using smoothing filtering to reduce noise on the spectral data obtained by scanning, and then standardizing the spectral data through Box-Cox transformation, and extracting features from the preprocessed spectral data. Feature extraction includes: dividing the spectral data into two-dimensional grids, dividing them into 100 two-dimensional grids with a side length of 1 cm, and marking each square in the grid to obtain the identification ID of each cube; using a peak detection algorithm to identify the characteristic peaks in each two-dimensional grid, and matching the corresponding characteristic peaks with the identification ID; extracting the spectral peak signals of glass beads with different refractive indices from a comparison database, and comparing the characteristic peaks one by one according to the identification ID, identifying the characteristic peaks with a spectral peak signal similarity of more than 90% in the comparison database as qualified characteristic peaks and extracting them, and calculating the spectral peak signal characteristic data based on the extracted characteristic peak data; The data analysis unit is connected to the data preprocessing unit and the spectrum database unit respectively, and is configured to use the Lasso regression algorithm to analyze the spectrum peak signal characteristic data and the data in the spectrum database to obtain the final glass bead content; The inversion unit is connected to the data preprocessing unit and the data analysis unit respectively, and is configured to determine the target data range of inversion according to the data in the spectral database; then use the spectral peak signal characteristic data to construct a spectral peak signal data dictionary; according to the principle of matching pursuit method, the data to be inverted are matched point by point with the characteristic data of the spectral peak signal data dictionary, and the data pairs with the greatest matching degree are extracted to form an inverted spectral peak signal data body; based on the spectral peak signal data dictionary, the inverted spectral peak signal data body and the comparison database, a deep learning network for inverting the reflection coefficient of the actual road surface marking is constructed, and the data sets in the comparison database and the spectral database are used to train the deep learning network to obtain a glass bead content prediction inversion model; the actual road surface marking area is input into the inversion model to obtain the reflection coefficient of the road surface marking after the final construction.

9. A computer device, characterized in that: The device comprises: processor; A memory storing a computer program, which, when executed by a processor, implements the method for detecting the glass bead content in a road marking as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method for detecting the glass bead content in a road marking as described in any one of claims 1 to 7 is implemented.

Images