License plate recognition method and system based on unmanned aerial vehicle adaptive shooting space modeling

Abstract

The invention relates to the technical field of data recognition, in particular to a license plate recognition method and system based on unmanned aerial vehicle adaptive shooting space modeling, and the method comprises the following steps: S10, constructing an adaptive shooting space model; s20, a camera carried by the unmanned aerial vehicle collects a license plate image within the range of the adaptive shooting space model; and S30, identifying the license plate number based on the acquired license plate image. The system comprises a self-adaptive shooting space model construction device used for constructing a self-adaptive shooting space model; the unmanned aerial vehicle is used for carrying a camera to acquire a license plate image within the range of the adaptive shooting space model; the image recognition device is used for recognizing the license plate number based on the collected license plate image. According to the invention, the camera of the unmanned aerial vehicle is limited in a specific shooting space model range, so that the quality of the acquired license plate image is high, the license plate recognition precision is greatly improved, and the license plate recognition efficiency is also improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data recognition, and in particular to a license plate recognition method and system based on unmanned aerial vehicle adaptive shooting space modeling. Background Art

[0002] Although people's awareness of traffic safety is gradually increasing, the phenomenon of randomly parking vehicles still exists. In response to the phenomenon of illegal parking, the current detection method is for traffic police to take photos on the spot to collect evidence, which is inefficient and has also caused the problem of "guerrilla warfare" between car owners and traffic police.

[0003] With the rapid development of drone technology, drone inspections have been widely used in many fields. In view of the low efficiency, high cost, and narrow monitoring area of ​​traditional inspections, which mainly rely on traffic police patrols and fixed camera capture, some scholars have proposed the idea of ​​using drones to inspect and take photos of illegally parked vehicles. For example, Wang Chen et al. proposed a method for urban road parking inspection based on drone images, which improves the all-weather inspection capability through illumination enhancement and deblurring algorithms, and improves YOLOv5 to improve detection accuracy. However, if the shooting distance is too far or the angle is inappropriate, the license plate will be blurred, deformed, or missing details, which will lead to low license plate recognition accuracy. Therefore, the existing methods still have certain limitations in adaptability and stability. Summary of the invention

[0004] The object of the present invention is to provide a license plate recognition method and system based on unmanned aerial vehicle adaptive shooting space modeling, so as to improve the efficiency and accuracy of license plate recognition.

[0005] In order to achieve the above-mentioned object of the invention, the embodiment of the present invention provides the following technical solutions:

[0006] A license plate recognition method based on drone adaptive shooting space modeling includes the following steps:

[0007] S10, constructing an adaptive shooting space model;

[0008] S20, the camera carried by the drone collects license plate images within the range of the adaptive shooting space model;

[0009] S30, recognizing the license plate number based on the collected license plate image.

[0010] Compared with the traditional method of using drones to collect license plate images at random locations, the above solution can greatly improve the quality of license plate images by limiting the drones to carry cameras to collect license plate images within a specific spatial range, thereby improving the accuracy of license plate recognition. Moreover, due to the high quality of license plate images, there is no need to collect them multiple times, thereby improving the efficiency of license plate image collection.

[0011] The S10 comprises the following steps:

[0012] S101, constructing an ellipsoid model as the adaptive shooting space model with the length direction of the road as the x-axis, the width direction of the road as the y-axis, and the direction perpendicular to the road plane as the z-axis;

[0013] The ellipsoid model is , a is the horizontal distance limit between the drone and the license plate; b is the lateral offset limit of the drone; c is the shooting height limit of the drone; (x i ,y i ,z i ) is the coordinate point of the drone position, (x0, y0, z0) is the coordinate point of the ellipsoid center, z i ≥z0, x1 and x2 are the closest and farthest horizontal shooting distances of the drone from the license plate, respectively.

[0014] Through comparative analysis of multiple models, the ellipsoid model is the most suitable as the shooting space model. That is, in the above scheme, by limiting the ellipsoid model as the adaptive shooting space model of the drone, the quality of the collected license plate image can be reliably guaranteed.

[0015] After S101, the method further includes step S102, in which a convex hull area with a license plate recognition accuracy higher than an accuracy threshold is fitted using a convex hull method, and the ellipsoid model is fitted using a minimum volume enclosing ellipsoid method.

[0016] In the above scheme, the ellipsoid model is fitted based on the license plate recognition accuracy, and the minimum volume enclosing ellipsoid method is used for fitting, which can reduce the range of the ellipsoid model as much as possible, and then reduce the occlusion of the camera by obstacles as much as possible, thereby improving the shooting efficiency.

[0017] After S102, the method further includes step S103 of correcting the ellipsoid model based on obstacles;

[0018] In S20, the camera carried by the drone collects the license plate image within the modified adaptive shooting space model.

[0019] There will inevitably be obstacles such as trees in the parking area. In the above scheme, after the shooting space model is initially formulated, the model is corrected based on the obstacles, and then the drone is placed within the corrected space range. This can eliminate the interference of obstacles and further improve the license plate image quality and shooting efficiency.

[0020] In S103, the process of correcting the ellipsoid model based on the obstacle includes:

[0021] The ray R(t) pointing from the drone position U to the license plate position P is defined as: Where t represents the distance the ray extends from the drone, and the direction vector It is expressed as: ;

[0022] Detect whether there is an obstacle intersecting with the ray R(t). If not, do not modify the ellipsoid model. If so, calculate the occlusion area of ​​the obstacle and define the occlusion area of ​​the obstacle as the intersection of the ray R(t) and the obstacle Q. i The set of intersection positions has , then the calculation formula of the modified shooting space model is: , O st Indicates the occluded area of ​​the obstacle.

[0023] a=6.23, b=5.15, c=6.24, and the coordinates of the center point are (6.69,0,0.05).

[0024] For the camera used in the experiment, the ellipsoid model defined in the above scheme is more optimal, and has the smallest spatial range while ensuring the shooting quality.

[0025] In S30, the YOLOv5-LPRNet model is used to recognize the license plate image.

[0026] The YOLO series of algorithms have fast detection speed and high accuracy, and the LPRNet license plate recognition algorithm has a highly lightweight network structure, which can meet the needs of limited hardware resources while ensuring high recognition accuracy. The above solution uses YOLOv5-LPRNet for license plate detection and recognition tasks, which can improve license plate recognition accuracy, occupy less resources, and have high efficiency.

[0027] A license plate recognition system based on UAV adaptive shooting space modeling, comprising:

[0028] An adaptive shooting space model building device, used to build an adaptive shooting space model;

[0029] A drone, used to carry a camera to collect license plate images within the range of the adaptive shooting space model;

[0030] The image recognition device is used to recognize the license plate number based on the collected license plate image.

[0031] The adaptive shooting space model is: , a is the horizontal distance limit between the drone and the license plate; b is the lateral offset limit of the drone; c is the shooting height limit of the drone; (x i ,y i ,z i ) is the coordinate point of the drone position, (x0, y0, z0) is the coordinate point of the ellipsoid center, z i≥z0, x1 and x2 are the closest and farthest horizontal shooting distances of the drone from the license plate, respectively.

[0032] The adaptive shooting space model construction device is also used to correct the adaptive shooting space model based on obstacles, and the drone is specifically used to carry a camera to collect license plate images within the range of the corrected adaptive shooting space model.

[0033] Compared with the prior art, the present invention aims to solve the problem of low license plate recognition accuracy caused by factors such as license plate tilt, blur, different scales and obstacle obstruction during the identification of illegally parked vehicles on the ground. An adaptive ellipsoid shooting space model that takes into account the relative distance between the drone and the license plate is constructed, so that the drone can flexibly adjust its position within the shooting space to optimize the shooting angle and imaging effect, so that the collected license plate image has high quality, thereby greatly improving the license plate recognition accuracy and efficiency.

[0034] Other technical advantages of the present invention are described in the following embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0036] Figure 1 It is a flow chart of the license plate recognition method based on drone adaptive shooting space modeling in the embodiment.

[0037] Figure 2 The flowchart of constructing the shooting space model in the embodiment is shown in FIG.

[0038] Figure 3 Schematic diagram of the ellipsoid model.

[0039] Figure 4 A schematic diagram of the grid.

[0040] Figure 5 This is a three-dimensional surface diagram of license plate recognition accuracy.

[0041] Figure 6 It is a two-dimensional surface diagram of license plate recognition accuracy at different heights.

[0042] Figure 7 Schematic diagram of the three-dimensional convex hull.

[0043] Figure 8 Schematic diagram of the minimum volume enclosing ellipsoid.

[0044] Fig. 9 Schematic diagram of the optimized ellipsoid. DETAILED DESCRIPTION

[0045] 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 only a part of the embodiments of the present invention, rather than all the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents the selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present invention.

[0046] See also Figure 1 The license plate recognition method based on drone adaptive shooting space modeling provided in this embodiment includes the following steps:

[0047] S10, constructing an adaptive shooting space model.

[0048] S20, the camera carried by the drone collects license plate images within the range of the adaptive shooting space model.

[0049] S30, recognizing the license plate number based on the collected license plate image.

[0050] At a closer distance, the drone's shooting height and left and right offset are smaller, and the license plate image is larger and clearer. As the horizontal distance increases, the shooting range gradually expands, but the proportion of the license plate in the image gradually decreases. When the distance continues to increase, the image details may not be enough to ensure high-precision recognition. Therefore, after exceeding a certain critical distance, in order to ensure that the license plate image has enough pixels, the shooting height and left and right offsets will tend to shrink, thereby meeting the accuracy requirements of license plate recognition. In other words, compared to arbitrarily controlling the flight height and distance of the drone, limiting the drone to a specific spatial range will help improve the quality of the license plate image collected by the drone's camera, and then improve the accuracy of license plate recognition. This specific spatial range is the shooting space, which refers to the construction of a three-dimensional space for license plate image collection with high recognition accuracy based on the location of the illegally parked vehicle.

[0051] See also Figure 2 In this embodiment, the above step S10 includes the following steps:

[0052] S101, constructing an ellipsoid as an adaptive shooting space model.

[0053] Although the cuboid can provide a large degree of freedom for shooting for the drone, it lacks effective constraints on the shooting angle, especially when shooting at the top or edge of the cuboid. The hemisphere provides a large degree of freedom in the vertical direction, which helps to adjust the height of the drone, but the constraints in the horizontal direction are weak, especially in the front and back and left and right directions, which can easily lead to image distortion. The cone performs well in the constraints of the shooting angle, but its boundaries are relatively simple, and the adjustment of the front and back and left and right directions is not flexible enough, making it difficult to provide balanced image quality at all shooting positions. Based on the above analysis, the ellipsoid is the most suitable choice. The three-dimensional space of the ellipsoid can provide sufficient shooting freedom, and its three axes can effectively control the horizontal distance, lateral offset and height between the drone and the license plate. At the same time, the ellipsoid has a shrinking trend at the center, which helps to optimize the shooting angle and image quality. Therefore, in this embodiment, the ellipsoid is used as an adaptive shooting space model.

[0054] The midpoint of one side of the road is set as the global coordinate system that describes the entire road environment to determine the position of vehicles and obstacles, with the length of the road as the x-axis, the width of the road as the y-axis, and the direction perpendicular to the road plane as the z-axis.

[0055] The center of the license plate is used as the local coordinate system to define the position of the drone relative to the license plate. In actual shooting, the drone shoots at a certain height from the license plate, so only the part z≥z0 is taken. Assume that the closest horizontal shooting distance between the drone and the license plate is x1, which is determined by the camera focus distance; the farthest horizontal distance is x2, which is determined by the number of pixels required for license plate recognition and the proportion of the license plate target in the image (if the proportion is too small, the target may not be detected). The ellipsoid is truncated at x1 and x2 respectively. The schematic diagram of the shooting space of the ellipsoid is as follows: Figure 3 As shown, the rectangular body on the left side of the figure represents a vehicle, and the black rectangular area in the rectangular body represents a license plate.

[0056] Assuming that the center point of the ellipsoid is O, the expression of the shooting space G is:

[0057] ;

[0058] Where a is the radius of the horizontal axis, corresponding to the horizontal distance limit between the drone and the license plate; b is the radius of the lateral axis, corresponding to the lateral offset limit of the drone; c is the radius of the vertical axis, corresponding to the shooting height limit of the drone; (x i ,y i ,z i ) is the coordinate point of the drone position. (x0, y0, z0) is the coordinate point of the ellipsoid center.

[0059] S102, using a convex hull method to fit a convex hull area where the license plate recognition accuracy is higher than an accuracy threshold, and using a minimum volume enclosing ellipsoid method to fit the ellipsoid model.

[0060] The identification algorithm for illegally parked vehicles requires not only high-precision detection capabilities, but also fast processing speed and real-time performance. In this embodiment, YOLOv5-LPRNet is used for license plate detection and recognition tasks. The license plate recognition accuracy can be expressed as: , where P r Indicates the license plate character recognition accuracy; NC A Indicates the total number of characters in the license plate; C t Indicates the number of correctly recognized characters. Let the accuracy threshold be R th , select the coordinate point set M whose recognition accuracy is greater than the threshold, .

[0061] In this embodiment, the three-dimensional convex hull method is used to construct the minimum convex hull containing the entire high recognition accuracy (i.e., the recognition accuracy is higher than the set value, and the recognition accuracy refers to the ratio of correctly recognized license plates) point set, effectively determine the outer boundary of the point set, and extract the distribution of the high recognition accuracy point set in space, providing a geometric boundary for subsequent ellipsoid fitting.

[0062] For example, suppose the set of shooting coordinate points with 100% recognition accuracy is , the convex hull calculation formula is:

[0063] ;

[0064] in, represents a convex combination, Represents the weight coefficient, which satisfies the conditions of being non-negative and summing to one.

[0065] Fitting the ellipsoid requires that the convex hull boundary points should be included while no other point sets should be introduced. Therefore, the minimum volume enclosing ellipsoid (MVEE) method is selected for fitting. MVEE is the minimum ellipsoid that contains the convex body. The calculation formula is:

[0066] ; ;

[0067] Where: Represents an ellipsoid, which is composed of a 3×3 positive definite matrix Q and the center To express it, the eigenvector and eigenvalue of Q are the semi-axis direction and semi-major axis length of the ellipsoid, and the semi-major axis direction and semi-major axis length refer to the direction and vector of the long, medium and short axes of the ellipsoid, respectively. Express Transpose. d is the scaling factor, and d / 1 determines the degree of scaling of the ellipsoid.

[0068] In order to verify the effectiveness of the ellipsoid shooting space model, actual data sets were collected and experimental analysis was performed. The goal of the experiment is to test whether the shooting space defined by the ellipsoid model can describe the relationship between the position coordinates of the drone and the license plate recognition accuracy in a real environment, and to ensure that the license plate recognition accuracy is optimized within the range defined by the shooting space model.

[0069] Since the shooting space without obstacles was established first, data collection was completed in an open parking lot. In this scenario, it is assumed that there are no interference factors such as weather and lighting, and only the spatial relationship between the drone position and the parking vehicle is considered to verify the universality and effectiveness of the shooting space model. During the shooting, the height of the drone, the horizontal distance from the license plate, and the lateral distance are controlled, and the drone camera is adaptive to always keep the license plate in the center of the imaging.

[0070] Table 1: Data collection equipment

[0071]

[0072] The parameters of the data acquisition equipment are shown in Table 1. The focus distance of the DJI MAVIC 3 main camera is 1 meter to infinity, so 1 meter is used as the starting horizontal shooting distance. Figure 4 As shown, a 0.5m×1m grid is divided every one meter (indicated by d in the figure) in front of the license plate. Multiple (for example, 7) images are collected for each grid. The grid corresponding to one x is called a group. Each group of data LPDS is recorded as: , then the “camera-license plate” distance dataset can be expressed as: , where n is the value of the horizontal distance x.

[0073] This experiment is based on Windows environment, using PyTorch framework, combined with CCPD public data set and license plate images collected by drones, and pre-trained the YOLOv5-LPRnet model. Finally, the YOLOv5-LPRnet model achieved 99.3% recognition accuracy. The pre-trained model was used to locate license plates and recognize characters on the CCLP data set (license plate images collected at various angles and heights), and the recognition accuracy of each shooting coordinate point was counted.

[0074] The coordinate points of the drone and the corresponding recognition accuracy values ​​are mapped into a three-dimensional surface graph and a two-dimensional plane graph, respectively. Figure 5 and Figure 6 shown. Figure 5The spatial distribution of license plate recognition accuracy is shown. The results show that the recognition accuracy reaches 100% in some areas, but as the horizontal distance increases (such as after x=11 meters), the recognition accuracy decreases significantly and the 100% accuracy area shrinks significantly. Figure 6 The distribution of license plate recognition accuracy at different heights (z=0.55, 2.55, 4.55, 6.55) is shown. From the comparison of different heights, it can be seen that as the height increases, the high-precision recognition area begins to shrink.

[0075] Depend on Figure 5 and 6 It is found that the points with 100% recognition accuracy show a certain regional distribution. The distribution of data points with 100% recognition accuracy on the x-axis is 1 meter to 12 meters, and the y and z axes also expand first and then move toward the central axis. Based on the coordinate point set with 100% recognition accuracy, the convex hull is calculated to obtain the three-dimensional convex hull, such as Figure 7 The convex hull boundary can be used as the boundary of the actual shooting space model in theory. The convex hull boundary points are fitted using the MVEE algorithm to obtain an approximate ellipsoid of the convex hull, as shown in Figure 8 shown.

[0076] The coverage rate is used as an indicator to measure the effect of the fitted ellipsoid approaching the three-dimensional convex hull. It means that half of the ellipsoid covers the three-dimensional convex hull volume v s and the three-dimensional convex hull volume v g The ratio is denoted as , . Divide the three-dimensional convex hull into n triangular pyramids, whose volume is recorded as v g , , a i is the base area of ​​the triangular pyramid, u i is the height of the triangular pyramid, covering volume v s Computed using Monte Carlo sampling method.

[0077] For the camera used in Table 1, the coordinates of the center point of the ellipsoid are initially fitted to be x0=6.69, y0=0, z0=0.05, the a, b and c axes are 6.23, 6.15 and 6.74 respectively, and the volume is 540.85. However, the ellipsoid model does not completely coincide with the outer boundary of the convex hull, especially in the c-axis direction, indicating that the fitting result does not fully capture the spatial characteristics of the scattered points, and there is still invalid space.

[0078] To improve the fitting accuracy, the semi-axis parameters of the ellipsoid were further optimized. The center point of the ellipsoid and the major semi-axis (a-axis) were kept unchanged, and the lengths of the minor semi-axis (b-axis) and c-axis were gradually reduced by 0.5 meters each time. The effect of the optimization was judged by observing the redundant space between the ellipsoid model and the convex hull and calculating the coverage rate until the redundant space between the ellipsoid model and the convex hull was close to 0.

[0079] The adjusted parameters are a=6.23, b=5.15, c=6.24, and the volume of the ellipsoid is 419.31. After optimization, the ellipsoid is truncated at x=1 and x=12. The truncated half of the ellipsoid is closely attached to the three-dimensional convex hull, reducing the introduction of invalid space. Fig. 9 As shown, the spatial adaptability and accuracy of the license plate recognition system in practical applications are improved.

[0080] There may be trees or other objects around the road where the vehicle is parked. Therefore, after the drone enters the shooting space, there may be obstacles at certain locations that block the shooting line of sight. Therefore, after the shooting space model is constructed, the shooting space model needs to be corrected based on the obstacles. That is, in the above step S10, step S103 may also be included to correct the fitted ellipsoid shooting space model based on the obstacles.

[0081] In this embodiment, ray casting is used to detect whether the ray intersects with any obstacle. If there is an intersection point, the line of sight is blocked; otherwise, the line of sight is clear. The ray R(t) pointing from the drone position U to the license plate position P can be expressed as: , where the direction vector It is expressed as: , t represents the distance the ray extends from the drone.

[0082] Use the Slab Method intersection detection algorithm to check whether the ray R(t) intersects with the obstacle O i Intersect. For the x / y / z axis, calculate the intersection point between the ray and the obstacle on that axis , , the calculation formula is:

[0083] ;

[0084] In the formula, , Respectively represent the minimum and maximum boundaries of the obstacle on this axis; Indicates the coordinates of the starting point of the ray on this axis; Represents the component of the direction vector on this axis.

[0085] if , indicating that the component of the ray's direction vector on a certain axis is negative, that is, the direction of movement on this axis is opposite to the positive direction of the coordinate axis. The originally calculated time of entering the obstacle may be greater than the time of leaving, which is inconsistent with the actual situation, because the entry time should be earlier than the exit time. Therefore, the two values ​​are swapped to ensure that the entry time is always less than or equal to the exit time. Comprehensively calculate the time t when the ray enters the obstacle enter and the time t to leave the obstacle exit, the calculation formula is: , then: That is, whether an obstacle blocks the ray is determined by calculating the time of entry and exit.

[0086] The occlusion area of ​​the obstacle is defined as the distance between the ray R(t) and the obstacle O i The set of intersection positions has , then the calculation formula of the modified shooting space model is: .

[0087] That is to say, if an obstacle blocks the ray, the intersection of the ray at each boundary of the obstacle is calculated based on the time when the ray enters and leaves each boundary of the obstacle, and the space relatively blocked by the obstacle is calculated based on the obtained intersection. Then, the blocked space is eliminated to obtain the corrected model of the photographable space. Taking the x-axis boundary as an example, the coordinates of the obstacle entering and leaving the boundary can be expressed as:

[0088] ;

[0089] ;

[0090] According to the above formula, the intersection points of each boundary can be obtained, and the occluded space can be obtained by connecting each intersection point.

[0091] This embodiment also provides a license plate recognition system based on drone adaptive shooting space modeling, including an adaptive shooting space model building device, a drone equipped with a camera, and an image recognition device.

[0092] The adaptive shooting space model construction device is used to construct an adaptive shooting space model. The adaptive shooting space model construction device is a software system, which is integrated into a computer and implements the above steps S101-S103.

[0093] The drone is used to carry a camera to collect license plate images within the range of the adaptive shooting space model. In practical applications, after the drone is placed within the range of the adaptive shooting space model, it is also necessary to detect whether there are obstacles blocking the space. If there are obstacles blocking the space, the adaptive shooting space model construction device will also correct the adaptive shooting space model, and then place the drone within the range of the corrected adaptive shooting space model to collect license plate images, so as to avoid the license plate image being unclear due to obstacles and affecting license plate recognition.

[0094] The image recognition device is used to recognize the license plate number based on the collected license plate image. The image recognition device is a software system, integrated into a computer, and performs the training steps and recognition steps of the YOLOv5-LPRNet model. Of course, the image recognition device can also use other models to recognize license plates.

[0095] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A license plate recognition method based on UAV adaptive shooting space modeling, characterized in that: The following steps are involved: S10, constructing an adaptive shooting space model; S20, the camera carried by the drone collects license plate images within the range of the adaptive shooting space model; S30, recognizing the license plate number based on the collected license plate image.

2. The license plate recognition method based on UAV adaptive shooting space modeling according to claim 1 is characterized in that: The S10 comprises the following steps: S101, constructing an ellipsoid model as the adaptive shooting space model with the length direction of the road as the x-axis, the width direction of the road as the y-axis, and the direction perpendicular to the road plane as the z-axis; The ellipsoid model is , a is the horizontal distance limit between the drone and the license plate; b is the lateral offset limit of the drone; c is the shooting height limit of the drone; (x i ,y i ,z i ) is the coordinate point of the drone position, (x0, y0, z0) is the coordinate point of the ellipsoid center, z i ≥z0, x1 and x2 are the closest and farthest horizontal shooting distances of the drone from the license plate, respectively.

3. The license plate recognition method based on UAV adaptive shooting space modeling according to claim 2 is characterized in that: After S101, the method further includes step S102, in which a convex hull area with a license plate recognition accuracy higher than an accuracy threshold is fitted using a convex hull method, and the ellipsoid model is fitted using a minimum volume enclosing ellipsoid method.

4. The license plate recognition method based on UAV adaptive shooting space modeling according to claim 3 is characterized in that: After S102, the method further includes step S103 of correcting the ellipsoid model based on obstacles; In S20, the camera carried by the drone collects the license plate image within the modified adaptive shooting space model.

5. The license plate recognition method based on UAV adaptive shooting space modeling according to claim 4 is characterized in that: In S103, the process of correcting the ellipsoid model based on the obstacle includes: The ray R(t) pointing from the drone position U to the license plate position P is defined as: , where t represents the distance the ray extends from the drone and the direction vector It is expressed as: ; Detect whether there is an obstacle intersecting with the ray R(t). If not, do not modify the ellipsoid model. If so, calculate the occlusion area of ​​the obstacle and define the occlusion area of ​​the obstacle as the intersection of the ray R(t) and the obstacle Q. i The set of intersection positions has , then the calculation formula of the modified shooting space model is: .

6. The license plate recognition method based on UAV adaptive shooting space modeling according to claim 2 is characterized in that: a=6.23, b=5.15, c=6.24, and the coordinates of the center point are (6.69,0,0.05).

7. The license plate recognition method based on UAV adaptive shooting space modeling according to claim 1 is characterized in that: In S30, the YOLOv5-LPRNet model is used to recognize the license plate image.

8. A license plate recognition system based on UAV adaptive shooting space modeling, characterized in that: include: An adaptive shooting space model building device, used to build an adaptive shooting space model; A drone, used to carry a camera to collect license plate images within the range of the adaptive shooting space model; The image recognition device is used to recognize the license plate number based on the collected license plate image.

9. The license plate recognition system based on UAV adaptive shooting space modeling according to claim 8 is characterized in that: The adaptive shooting space model is: , a is the horizontal distance limit between the drone and the license plate; b is the lateral offset limit of the drone; c is the shooting height limit of the drone; (x i ,y i ,z i ) is the coordinate point of the drone position, (x0, y0, z0) is the coordinate point of the ellipsoid center, z i ≥z0, x1 and x2 are the closest and farthest horizontal shooting distances of the drone from the license plate, respectively.

10. The license plate recognition system based on UAV adaptive shooting space modeling according to claim 8 is characterized in that: The adaptive shooting space model construction device is also used to correct the adaptive shooting space model based on obstacles, and the drone is specifically used to carry a camera to collect license plate images within the range of the corrected adaptive shooting space model.

Images