Evaluation method for calculating road coverage based on roadside surveillance camera image recognition

Abstract

The present application belongs to the technical field of intelligent transportation systems, and discloses an evaluation method for calculating road coverage based on image recognition of roadside monitoring cameras, comprising the following steps: Step 1: obtaining road parameters of a road section to be evaluated; Step 2: inputting the road parameters into CARSIM simulation software, setting different monitoring camera parameters and road geometry conditions, simulating the monitoring scene through the CARSIM simulation software, and obtaining a coverage area image; Step 3: performing grayscale processing, color space conversion, image binarization and denoising processing on the simulated coverage area image in sequence, and extracting the road area; Step 4: using affine transformation technology to calculate the actual coverage area of ​​the road area in the coverage area image, and finally obtaining the road coverage rate.

Description

Technical Field

[0001] The present application belongs to the technical field of intelligent transportation systems, and specifically relates to an evaluation method for calculating road coverage based on image recognition of roadside monitoring cameras. Background Art

[0002] With the development of intelligent transportation systems, roadside surveillance cameras play a vital role in traffic management and safety monitoring. These cameras collect road images in real time to help traffic management departments identify and handle traffic accidents, traffic violations, road blockages and other issues in a timely manner, thereby effectively improving road traffic efficiency and traffic safety. However, with the complexity of road structures, how to achieve maximum coverage of road areas through the reasonable deployment of surveillance cameras has become an important technical challenge.

[0003] The deployment of surveillance cameras is mainly based on empirical rules and simple geometric analysis, which makes it difficult to accurately evaluate the actual coverage of the cameras. Factors such as the geometry of the road, the installation height and pitch angle of the camera will have a significant impact on the monitoring range. Due to the complex combination of these factors, traditional deployment methods are difficult to provide accurate coverage predictions, which in turn affects the overall effectiveness of the monitoring system.

[0004] Therefore, how to use advanced image recognition technology and simulation tools to accurately calculate the coverage of surveillance cameras under different road conditions has become a technical problem that needs to be solved urgently in this field. Currently, there is no research on this technology. Summary of the invention

[0005] The purpose of this application is to solve the problems of the prior art and to provide an evaluation method for calculating road coverage based on roadside monitoring camera image recognition.

[0006] In order to solve the technical problem, the technical solution of the present application is: an evaluation method for calculating road coverage based on roadside monitoring camera image recognition, comprising the following steps:

[0007] Step 1: Obtain the road parameters of the road section to be evaluated;

[0008] Step 2: Input the road parameters into the CARSIM simulation software, set different monitoring camera parameters and road geometry conditions, simulate the monitoring scene through the CARSIM simulation software, and obtain the coverage area image;

[0009] Step 3: The simulated coverage area image is subjected to grayscale processing, color space conversion, image binarization and denoising in sequence to extract the road area;

[0010] Step 4: Use affine transformation technology to calculate the actual coverage area of ​​the road area in the coverage area image, and finally obtain the road coverage rate.

[0011] Preferably, the road parameters in step 1 include road width, curve radius and road length.

[0012] Preferably, the monitoring camera parameters in step 2 include height, focal length, field of view angle and pitch angle, the pitch angle includes 20°, 35°, 50° or 65°, the height, focal length and field of view angle are set according to standard monitoring specifications, and the road geometry conditions include straight segments and curve segments with different curve radii.

[0013] Preferably, step 2 specifically includes: inputting the road parameters into the CARSIM simulation software, setting different monitoring camera parameters and road geometry conditions, using the CARSIM simulation software to model the road and the monitoring camera, placing the monitoring camera at the starting point of the road model, and adjusting the pitch angle of the monitoring camera to simulate different monitoring viewing angles to obtain a color coverage area image.

[0014] Preferably, the step 3 is specifically:

[0015] Step 3-1: Convert the simulated color coverage area image into a grayscale image, and use the weighted average method to weight the red, green, and blue channels in the ratio of 0.3, 0.59, and 0.11;

[0016] Step 3-2: Convert the grayscale image to HSV color space, define the color range of the road area, and create a mask through threshold segmentation method;

[0017] Step 3-3: Perform morphological processing on the mask and then extract the road area, wherein the morphological processing includes opening and closing operations to eliminate noise and fill small holes.

[0018] Preferably, the step 4 is specifically:

[0019] Step 4-1: Select reference points with known positions in the coverage area image, and calculate the affine transformation matrix through these reference points to convert the image coordinates into actual physical coordinates;

[0020] Step 4-2: Apply the affine transformation matrix to obtain the road area image in actual physical coordinates;

[0021] Step 4-3: Calculate the actual area of ​​each pixel in the road area image, combine the area of ​​the reference point with known position, and calculate the actual area of ​​each pixel by the corresponding number of pixels in the mask;

[0022] Step 4-4: Count the total number of pixels in the road area and multiply it by the actual area of ​​each pixel to obtain the actual coverage area, thereby obtaining the road coverage rate.

[0023] Preferably, the reference point in step 4-1 is a rectangular object, or a rectangle is set as the reference point in the CARSIM simulation software.

[0024] Preferably, the affine transformation matrix is:

[0025]

[0026] Where:

[0027] is the focal length of the surveillance camera, m;

[0028] is the field of view of the surveillance camera, rad;

[0029] is the pitch angle of the surveillance camera, rad;

[0030] is the height of the surveillance camera, m;

[0031] is the scale factor;

[0032] is the coordinate of a point in the image coordinate system;

[0033] is the coordinate of a point in the actual physical coordinate system.

[0034] Preferably, the calculation model of the actual coverage area is as follows:

[0035]

[0036] Where:

[0037] is the curve radius of the road, m;

[0038] is the pitch angle of the surveillance camera, rad.

[0039] Preferably, the calculation model of the actual coverage area is as follows:

[0040] Where:

[0041] is the quadratic term of the curve radius of the road, m 2 ;

[0042] is the quadratic term of the pitch angle of the surveillance camera, rad 2 ;

[0043] is the interaction term between curve radius and pitch angle.

[0044] Compared with the prior art, the advantages of this application are:

[0045] (1) This application proposes an evaluation method for calculating road coverage based on image recognition of roadside surveillance cameras. Through simulation technology, image processing technology and affine transformation technology, the coverage of the road area is accurately calculated, providing a scientific basis for the optimal layout of surveillance cameras. The method is fast and efficient, and can comprehensively consider multi-dimensional factors, such as the pitch angle of the surveillance camera and the curve radius of the road, and ensure the reliability of the evaluation results through a high-fitting regression model.

[0046] (2) This application obtains the geometric parameters of the road to be evaluated, including road width, curve radius, and installation parameters of the monitoring camera, through a fast, simple, and accurate evaluation of the road coverage rate. These parameters are input into the CARSIM simulation software to simulate the coverage scenarios under different monitoring camera pitch angles and road conditions. The road area is extracted by performing image processing on the coverage area image generated by the CARSIM simulation software, and the actual coverage area is accurately calculated using affine transformation technology to obtain an accurate coverage rate prediction, thereby improving the overall effect of the monitoring system.

[0047] (3) This application uses two actual coverage area calculation models to achieve accurate analysis and evaluation of road surveillance camera coverage. The linear regression model is used to simply and clearly evaluate the linear impact of the road curve radius and the pitch angle of the surveillance camera on the coverage area. This model shows that an increase in the curve radius or an increase in the pitch angle of the surveillance camera will lead to a decrease in the coverage area, but the two effects are relatively independent, which is convenient for preliminary optimization of the layout of surveillance cameras.

[0048] (4) This application further considers the quadratic effect and interaction term between curve radius and pitch angle through a polynomial regression model; this multi-dimensional evaluation method reveals the complex relationship between the two, that is, changes in curve radius will significantly affect the impact of pitch angle on coverage area, and vice versa; this model can more accurately predict the coverage effect in practical applications, especially under complex road conditions; the introduction of the polynomial regression model enables the evaluation to adapt to more actual working conditions and improves the accuracy and rationality of surveillance camera layout; combined with image processing and affine transformation, it ensures high-precision calculation of the monitoring area and provides a strong theoretical basis and practical guidance for the design and optimization of road traffic management systems; this application realizes a quick and simple evaluation of road surveillance camera coverage and can significantly improve the safety and management efficiency of road traffic. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1This is a flow chart of the evaluation method for calculating road coverage based on roadside surveillance camera image recognition in this application;

[0050] Figure 2 The color coverage area image in Example 1 of the present application;

[0051] Figure 3 The grayscale image is obtained by grayscale processing the color coverage area image in Example 1 of the present application;

[0052] Figure 4 The image is the image after the grayscale image is subjected to noise reduction processing in Example 1 of the present application;

[0053] Figure 5 This is a schematic diagram of performing affine transformation on the extracted road area in Example 1 of the present application;

[0054] Figure 6 This is the monitoring image of the actual engineering case in Example 2 of this application;

[0055] Figure 7 The coverage area image obtained by using the method of the present application in Example 2 of the present application;

[0056] Figure 8 This is the coverage area image obtained using the traditional method in Example 2 of the present application. DETAILED DESCRIPTION

[0057] The present application is described in detail below in conjunction with the accompanying drawings and specific embodiments, but the present application is not limited to these embodiments. The present application covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present application. In order to make the public have a thorough understanding of the present application, specific details are described in detail in the following embodiments of the present application, and those skilled in the art can fully understand the present application without the description of these details.

[0058] like Figure 1 As shown, the present application discloses a method for evaluating road coverage based on roadside surveillance camera image recognition, comprising the following steps:

[0059] Step 1: Obtain the road parameters of the road section to be evaluated;

[0060] Step 2: Input the road parameters into the CARSIM simulation software, set different monitoring camera parameters and road geometry conditions, simulate the monitoring scene through the CARSIM simulation software, and obtain the coverage area image;

[0061] Step 3: The simulated coverage area image is subjected to grayscale processing, color space conversion, image binarization and denoising in sequence to extract the road area;

[0062] Step 4: Use affine transformation technology to calculate the actual coverage area of ​​the road area in the coverage area image, and finally obtain the road coverage rate.

[0063] Preferably, the road parameters in step 1 include road width, curve radius and road length.

[0064] Preferably, the monitoring camera parameters in step 2 include height, focal length, field of view angle and pitch angle, the pitch angle includes 20°, 35°, 50° or 65°, the height, focal length and field of view angle are set according to standard monitoring specifications, and the road geometry conditions include straight segments and curve segments with different curve radii.

[0065] Preferably, step 2 specifically includes: inputting the road parameters into the CARSIM simulation software, setting different monitoring camera parameters and road geometry conditions, using the CARSIM simulation software to model the road and the monitoring camera, placing the monitoring camera at the starting point of the road model, and adjusting the pitch angle of the monitoring camera to simulate different monitoring viewing angles to obtain a color coverage area image.

[0066] The CARSIM simulation software is existing software, and the specific operation and usage methods will not be repeated in this application.

[0067] Preferably, the step 3 is specifically:

[0068] Step 3-1: Convert the simulated color coverage area image into a grayscale image, and use the weighted average method to weight the red, green, and blue channels in the ratio of 0.3, 0.59, and 0.11;

[0069] Step 3-2: Convert the grayscale image to HSV color space, define the color range of the road area, and create a mask through threshold segmentation method;

[0070] Step 3-3: Perform morphological processing on the mask and then extract the road area, wherein the morphological processing includes opening and closing operations to eliminate noise and fill small holes.

[0071] Preferably, the step 4 is specifically:

[0072] Step 4-1: Select reference points with known positions in the coverage area image, and calculate the affine transformation matrix through these reference points to convert the image coordinates into actual physical coordinates;

[0073] Step 4-2: Apply the affine transformation matrix to obtain the road area image in actual physical coordinates;

[0074] Step 4-3: Calculate the actual area of ​​each pixel in the road area image, combine the area of ​​the reference point with known position, and calculate the actual area of ​​each pixel by the corresponding number of pixels in the mask;

[0075] Step 4-4: Count the total number of pixels in the road area and multiply it by the actual area of ​​each pixel to obtain the actual coverage area, thereby obtaining the road coverage rate.

[0076] Preferably, the reference point in step 4-1 is a rectangular object, or a rectangle is set as the reference point in the CARSIM simulation software.

[0077] Preferably, the affine transformation matrix is:

[0078]

[0079] Where:

[0080] is the focal length of the surveillance camera, m;

[0081] is the field of view of the surveillance camera, rad;

[0082] is the pitch angle of the surveillance camera, rad;

[0083] is the height of the surveillance camera, m;

[0084] is the scale factor;

[0085] is the coordinate of a point in the image coordinate system;

[0086] is the coordinate of a point in the actual physical coordinate system.

[0087] Preferably, the calculation model of the actual coverage area is as follows:

[0088]

[0089] Where:

[0090] is the curve radius of the road, m;

[0091] is the pitch angle of the surveillance camera, rad.

[0092] Preferably, the calculation model of the actual coverage area is as follows:

[0093] Where:

[0094] is the quadratic term of the curve radius of the road, m 2 ;

[0095] is the quadratic term of the pitch angle of the surveillance camera, rad 2 ;

[0096] is the interaction term between curve radius and pitch angle.

[0097] Example 1

[0098] like Figure 1 FIG. 1 is a flowchart of a method for evaluating road coverage based on roadside surveillance camera image recognition proposed in an embodiment of the present application. The method includes the following steps:

[0099] Step 1: Obtain the road parameters of the road section to be evaluated.

[0100] In the implementation of this application, it is first necessary to obtain the road parameters of the road section to be evaluated, including road width, curve radius, road length, etc. These parameters provide basic data for subsequent coverage calculation and risk assessment.

[0101] Step 2: Input the road parameters into the CARSIM simulation software, set different monitoring camera parameters and road geometry conditions, simulate the monitoring scene through the simulation software, and obtain the coverage area image.

[0102] This application uses simulation software to model roads and surveillance cameras, based on a 7.5-meter-wide two-lane road model. A 7.5-meter-wide two-lane road is created in the simulation software. The road model includes straight segments and curve segments with different curve radii. A surveillance camera is placed at the starting point of the road model, and the pitch angle of the surveillance camera is adjusted to simulate different surveillance viewing angles. Parameters such as the height, focal length, and field of view of the surveillance camera are set according to standard monitoring specifications to ensure the accuracy of the simulation process.

[0103] like Figure 2 As shown, a color coverage area image is obtained by using CARSIM simulation software to model the road and the surveillance camera. A surveillance camera is placed at the starting point of the road model, and the pitch angle of the surveillance camera is adjusted to simulate different monitoring perspectives to obtain a color coverage area image.

[0104] Step 3: Use image processing technology to grayscale, convert color space, binarize and denoise the simulated image to extract the road area.

[0105] like Figure 3As shown in the figure, the grayscale image after grayscale processing of the color coverage area image is obtained. In the image acquisition and preprocessing stage, the image under the perspective of the simulated surveillance camera is first obtained. These images usually contain rich color information, but may be interfered by the background. In order to reduce this interference, the image is first grayscaled. In this process, the red, green and blue color channels of the color image are weighted averaged according to a preset weighting ratio to generate a single-channel grayscale image. The generation of grayscale images can effectively reduce the interference caused by color information, thereby improving the accuracy of subsequent image segmentation.

[0106] In the color range definition and mask creation steps, the grayscale image is converted to the HSV color space. In the HSV color space, color information and brightness information are separated, making color segmentation more accurate. By defining a specific color range of the road area in the HSV space and using the threshold segmentation method to create a mask, the road area can be effectively extracted from the image. The execution of this step can significantly filter out most of the background noise, laying a solid foundation for subsequent image processing.

[0107] like Figure 4 As shown in the figure, the image after denoising the grayscale image. In the image denoising step, noise points or small holes may appear in the created mask. These small defects will affect the integrity of the mask. Therefore, morphological processing is used to further optimize the mask. In the specific operation, the morphological opening operation is used to eliminate the small noise points in the mask, and the closing operation is used to fill the small holes. These morphological operations can significantly improve the quality of the mask, making the extracted road area more coherent and complete.

[0108] The opening operation and closing operation are existing methods and will not be described in detail in this application.

[0109] In the road area segmentation step, the processed mask is applied to the original image to make the road area more prominent and obvious in the image. Specifically, by superimposing the previously generated mask layer on the original image, this operation effectively isolates the road area from the background area, making the road area present a clearer and more defined boundary in the image.

[0110] Step 4: Use affine transformation technology to calculate the actual coverage area of ​​the road area in the image and finally obtain the coverage rate.

[0111] In the embodiment of the present application, the actual coverage area is determined by the following steps:

[0112] B1. Figure 5 As shown in FIG. 1 , it is a schematic diagram of performing affine transformation on the extracted road area. Figure 5 (a) is the extracted road area. Figure 5(b) is the area after the affine transformation of the extracted road area; select reference points with known positions in the coverage area image, and calculate the affine transformation matrix through these reference points to convert the image coordinates into actual physical coordinates. Specifically, it is first necessary to select reference points with known positions in the coverage area image. The reference points are usually road markings or other features with fixed positions, which have clear positions in both the image and the actual physical environment. Through these reference points, the affine transformation matrix can be calculated. The affine transformation matrix is ​​a mathematical tool used to convert the pixel coordinate system in the image into the corresponding physical coordinate system.

[0113] B2. Apply affine transformation to obtain the road area image under the actual physical coordinates. Specifically, after calculating the affine transformation matrix, apply the matrix to the entire image to generate the road area image under the actual physical coordinates. Through the affine transformation, all pixels in the image are remapped into a regular rectangular area, which represents the physical characteristics of the actual road. This step not only simplifies the irregularly shaped road area, but also ensures that each pixel corresponds accurately to the actual road physical space, thereby achieving accurate calculation of complex road areas.

[0114] B3. Calculate the actual area of ​​each pixel, and combine it with the known area of ​​the reference point to calculate the actual area of ​​each pixel by the corresponding number of pixels in the mask. Specifically, the area of ​​each pixel in the physical space after the affine transformation is uniform, so the actual area of ​​each pixel can be inferred by calculating the number of pixels in the reference point area in the transformed image and combining it with the known actual area of ​​the area.

[0115] B4. Count the total number of pixels in the road area and multiply it by the actual area of ​​each pixel to obtain the actual coverage area of ​​the road.

[0116] In order to make the simulation model simulate the actual scene more accurately, the model needs to be calibrated with data. Specifically, a 7.5-meter-wide two-lane road model was constructed in the CARSIM simulation software, including straight sections and curves with different curve radii. In order to study the impact of the pitch angle of the surveillance camera on the coverage rate, four common pitch angles were set in the experiment: 20°, 35°, 50°, and 65°. These angles represent the common surveillance camera configurations in actual road monitoring, covering the monitoring viewing angle range from low to high. In order to simulate the impact of different road geometry conditions on the monitoring effect, the experiment selected multiple different road curve radii, including 250 meters, 500 meters, 750 meters, 1000 meters, 2500 meters, 5000 meters and straight sections. By setting these different curve radii, the impact of road geometry on the coverage of the surveillance camera can be comprehensively analyzed. After setting the pitch angle of the surveillance camera and the curve radius of the road, the simulation software was used to simulate the viewing angle of the surveillance camera and capture road area images under different combinations of conditions. The actual road coverage area data was extracted from the acquired images, and statistics and data analysis were performed. The results are shown in Table 1 below:

[0117]

[0118] To better understand the pitch angle of the surveillance camera ( ) and the curve radius of the road ( ) on the actual coverage area, this application performed a regression analysis to quantify the contribution of each variable to the coverage area and reveal the trend of coverage under different conditions. Before performing the regression analysis, this application performed a comprehensive preprocessing of the experimental data. First, the data was normalized for different measurement units of curve radius and pitch angle (meters and degrees, respectively) to ensure the comparability of each variable in the model. The normalization process adjusted the mean of each variable to 0 and the standard deviation to 1.

[0119] This application firstly applies a linear regression model to explore the linear relationship between the actual coverage area and the curve radius and pitch angle. The experimental data is fitted by the least squares method to obtain the preliminary regression coefficient.

[0120]

[0121] In the formula, is the curve radius of the road, m; is the pitch angle of the surveillance camera, rad.

[0122] However, since the relationship between the curve radius and the pitch angle and the actual coverage area may be nonlinear, and there may be interaction between the two, in order to improve the fit of the model, this application introduces a polynomial regression model.

[0123] In the polynomial regression model, the quadratic terms of curve radius and pitch angle and their interaction terms are considered. This is because as the curve radius changes, the actual coverage area may show a trend of first increasing and then decreasing, and at a larger pitch angle, the effect of the curve radius on the actual coverage area may change. By introducing quadratic terms and interaction terms, the model can capture these complex relationships more accurately. The mathematical expression of the polynomial regression model is as follows:

[0124]

[0125] In the formula, is the quadratic term of the curve radius of the road, m 2 ; is the quadratic term of the pitch angle of the surveillance camera, rad 2 ; is the interaction term between curve radius and pitch angle.

[0126] The actual coverage area is a linear combination of each coefficient and the curve radius, the square of the curve radius, the pitch angle, the square of the pitch angle, and the interaction between the curve radius and the pitch angle.

[0127] The mean square error (MSE) of the polynomial regression model was calculated to be 2010.493, and the coefficient of determination (R²) was 0.971, indicating that the model fit the experimental data very well.

[0128] The evaluation of the polynomial regression model shows that the model has high accuracy. The mean square error (MSE) represents the average square difference between the predicted value and the actual value. The MSE value is 2010.493, indicating that the model prediction accuracy is high. The coefficient of determination (R²) is 0.971, indicating that the model can well explain the variability of the data and has a high degree of fit. These results prove the necessity and effectiveness of introducing quadratic terms and interaction terms, ensuring the reliability of the model in practical applications.

[0129] Through the above polynomial regression analysis, the conclusions and models obtained in this application provide a scientific basis for the deployment of road monitoring cameras, which helps traffic management departments to more effectively plan and optimize monitoring systems, thereby improving road traffic safety and management efficiency. Future research can further explore the impact of other variables (such as camera installation height, different weather conditions, etc.) on coverage, and provide a more comprehensive optimization solution for intelligent transportation systems. This solution will provide an important reference for the design and implementation of intelligent transportation systems and promote the continuous advancement and improvement of traffic management technology.

[0130] Example 2

[0131] In order to further verify the effectiveness and superiority of the method described in this application, the method of this application is applied to actual working conditions and compared with the existing traditional methods. The Yuhuang toll station is selected as the experimental scene, and the specific implementation process is as follows by comparing the real shot images with the simulated processed images:

[0132] like Figure 6 As shown, it is a monitoring image of an actual engineering case, which shows a real-life picture taken by a monitoring camera at the Yuhuang Toll Station. The road structure in this area is complex, including multiple lanes and curves. According to the method steps of this application, the road parameters of the Yuhuang Toll Station are obtained, including road width, curve radius, and road length. These parameters are input into the CARSIM simulation software, the parameters of the monitoring camera (height, focal length, field of view, and pitch angle) are set, and different road geometry conditions are simulated to generate an image of the coverage area. Subsequently, image processing technology is applied to grayscale, color space conversion, binarization, and denoising of the simulated image, as shown in FIG. Figure 7 As shown in the figure, the coverage area image obtained by the method of the present application is extracted from the road area, and the actual coverage area is 1428.67m 2 , which is 96% consistent with the actual area. Figure 8 The figure shows the coverage area image obtained by the traditional method, and the coverage area image processed by the traditional affine transformation method. It can be observed that the traditional method is not accurate enough in extracting the coverage area under complex road structures and multi-angle surveillance camera perspectives. The calculated degree of agreement with the actual real area is 86%, which has obvious errors and cannot effectively reflect the actual road coverage.

[0133] This application proposes an evaluation method for calculating road coverage based on image recognition of roadside surveillance cameras. Through simulation technology, image processing technology and affine transformation technology, accurate coverage calculation of road areas is performed, providing a scientific basis for the optimal layout of surveillance cameras. This method is fast and efficient, and can comprehensively consider multi-dimensional factors, such as the pitch angle of the surveillance camera and the radius of the road curve, and ensure the reliability of the evaluation results through a high-fitting regression model.

[0134] This application obtains the geometric parameters of the road to be evaluated, including road width, curve radius, and installation parameters of the monitoring camera, through a fast, simple and accurate evaluation of the road coverage rate, and inputs these parameters into the CARSIM simulation software to simulate the coverage scenarios under different monitoring camera pitch angles and road conditions. The coverage area images generated by the CARSIM simulation software are processed, the road area is extracted, and the affine transformation technology is used to accurately calculate the actual coverage area to obtain an accurate coverage rate prediction, thereby improving the overall effect of the monitoring system.

[0135] This application realizes accurate analysis and evaluation of the coverage rate of road surveillance cameras through two actual coverage area calculation models; through the linear regression model, the linear impact of road curve radius and surveillance camera pitch angle on the coverage area is simply and clearly evaluated. This model shows that an increase in the curve radius or an increase in the pitch angle of the surveillance camera will lead to a decrease in the coverage area, but the impact of the two is relatively independent, which is convenient for the preliminary optimization of the layout of surveillance cameras.

[0136] This application further considers the quadratic effect and interaction term between curve radius and pitch angle through a polynomial regression model; this multi-dimensional evaluation method reveals the complex relationship between the two, that is, changes in curve radius will significantly affect the impact of pitch angle on coverage area, and vice versa; this model can more accurately predict the coverage effect in practical applications, especially under complex road conditions; the introduction of the polynomial regression model enables the evaluation to adapt to more actual working conditions, and improves the accuracy and rationality of the deployment of surveillance cameras; combined with image processing and affine transformation, it ensures high-precision calculation of the monitoring area, and provides a strong theoretical basis and practical guidance for the design and optimization of road traffic management systems; this application realizes a quick and simple evaluation of road surveillance camera coverage, and can significantly improve the safety and management efficiency of road traffic.

[0137] The preferred implementation modes of the present application have been described in detail above, but the present application is not limited to the above implementation modes, and various changes can be made within the knowledge scope of ordinary technicians in this field without departing from the purpose of the present application.

[0138] Many other changes and modifications can be made without departing from the concept and scope of the present application.It should be understood that the present application is not limited to specific embodiments, and the scope of the present application is defined by the appended claims.

Claims

1. An evaluation method for calculating road coverage based on image recognition of roadside surveillance cameras, characterized in that: The following steps are involved: Step 1: Obtain the road parameters of the road section to be evaluated; Step 2: Input the road parameters into the CARSIM simulation software, set different monitoring camera parameters and road geometry conditions, simulate the monitoring scene through the CARSIM simulation software, and obtain a coverage area image; Input the road parameters into the CARSIM simulation software, set different monitoring camera parameters and road geometry conditions, use the CARSIM simulation software to model the road and the monitoring camera, place the monitoring camera at the starting point of the road model, and adjust the pitch angle of the monitoring camera to simulate different monitoring viewing angles to obtain a color coverage area image; Step 3: The simulated coverage area image is subjected to grayscale processing, color space conversion, image binarization and denoising in sequence to extract the road area; Step 3-1: Convert the simulated color coverage area image into a grayscale image, and use the weighted average method to weight the red, green, and blue channels in the ratio of 0.3, 0.59, and 0.11; Step 3-2: Convert the grayscale image to HSV color space, define the color range of the road area, and create a mask through threshold segmentation method; Step 3-3: Perform morphological processing on the mask and then extract the road area, wherein the morphological processing includes opening and closing operations to eliminate noise and fill small holes; Step 4: Use affine transformation technology to calculate the actual coverage area of ​​the road area in the coverage area image, and finally obtain the road coverage rate.

2. The method for evaluating road coverage based on roadside surveillance camera image recognition according to claim 1 is characterized in that: The road parameters in step 1 include road width, curve radius and road length.

3. The method for evaluating road coverage based on roadside surveillance camera image recognition according to claim 1 is characterized in that: The monitoring camera parameters in step 2 include height, focal length, field of view angle and pitch angle, the pitch angle includes 20°, 35°, 50° or 65°, the height, focal length and field of view angle are set according to standard monitoring specifications, and the road geometry conditions include straight segments and curve segments with different curve radii.

4. The method for evaluating road coverage based on roadside surveillance camera image recognition according to claim 1 is characterized in that: The step 4 is specifically as follows: Step 4-1: Select reference points with known positions in the coverage area image, and calculate the affine transformation matrix through these reference points to convert the image coordinates into actual physical coordinates; Step 4-2: Apply the affine transformation matrix to obtain the road area image in actual physical coordinates; Step 4-3: Calculate the actual area of ​​each pixel in the road area image, combine the area of ​​the reference point with known position, and calculate the actual area of ​​each pixel by the corresponding number of pixels in the mask; Step 4-4: Count the total number of pixels in the road area and multiply it by the actual area of ​​each pixel to obtain the actual coverage area, thereby obtaining the road coverage rate.

5. The method for evaluating road coverage based on roadside surveillance camera image recognition according to claim 4 is characterized in that: In step 4-1, the reference point is a rectangular object, or a rectangle is set as the reference point in the CARSIM simulation software.

6. The method for evaluating road coverage based on roadside surveillance camera image recognition according to claim 4 is characterized in that: The affine transformation matrix is: ； Where: is the focal length of the surveillance camera, m; is the field of view of the surveillance camera, rad; is the pitch angle of the surveillance camera, rad; is the height of the surveillance camera, m; is the scale factor; is the coordinate of a point in the image coordinate system; is the coordinate of a point in the actual physical coordinate system.

7. The method for evaluating road coverage based on roadside surveillance camera image recognition according to claim 4 is characterized in that: The calculation model of the actual coverage area is as follows: ； Where: is the curve radius of the road, m; is the pitch angle of the surveillance camera, rad.

8. The method for evaluating road coverage based on roadside surveillance camera image recognition according to claim 4 is characterized in that: The calculation model of the actual coverage area is as follows: ； Where: is the quadratic term of the curve radius of the road, m 2 ; is the quadratic term of the pitch angle of the surveillance camera, rad 2 ; is the interaction term between curve radius and pitch angle.

Images