Accident restoration method and device, electronic equipment and storage medium

Abstract

The invention provides an accident restoration method and device, electronic equipment and a storage medium, and the method comprises the steps: determining a flight path parameter of aerial photographing equipment according to a test image sequence of a test accident obtained by the aerial photographing equipment in a test scene; in an actual scene, acquiring a first image sequence shot by the aerial photographing equipment in a flight process of the aerial photographing equipment according to the flight path parameter; wherein the first image sequence comprises multiple frames of first images, and the next frame of first image is obtained by shooting after shooting parameters of the previous frame of first image are optimized; and displaying a first accident scene obtained by performing reconstruction simulation according to the first image sequence. Through efficient data acquisition of the unmanned aerial vehicle and accurate analysis of a computer vision technology, three-dimensional reconstruction and kinetic analysis of an accident scene can be efficiently completed, and a scientific basis is provided for accident liability affirmation.

Description

Technical Field

[0001] The present application relates to the field of accident restoration technology, and more specifically, to an accident restoration method, device, electronic device and storage medium. Background Art

[0002] The most important task of accident reconstruction is to use the remaining driving marks on the road at the accident scene, the collision marks on the vehicle body and the stopped position of the vehicle to estimate the vehicle's speed, acceleration, heading angle during the collision and the vehicle's driving trajectory after the collision, so as to infer the cause of the traffic accident, determine the responsibility for the accident and take corresponding measures.

[0003] Traditional accident scene investigation mainly relies on manual measurement and photographic evidence collection. This method has problems such as insufficient manual operation accuracy and time-consuming. Summary of the invention

[0004] In view of this, the purpose of the present application is to provide a method, device, electronic device and storage medium for accident restoration to overcome the problems in the prior art.

[0005] In a first aspect, an embodiment of the present application provides a method for accident recovery, the method comprising: In a test scenario, determining a flight path parameter of the aerial photography device according to a test image sequence of a test accident acquired by the aerial photography device; In an actual scenario, when the aerial photography device flies according to the flight path parameters, a first image sequence shot by the aerial photography device is obtained; wherein the first image sequence includes a plurality of first image frames, and a subsequent first image frame is shot after optimizing the shooting parameters of a previous first image frame; The first accident scene obtained by reconstructing and simulating the first image sequence is displayed.

[0006] In some technical solutions of the present application, in the test scenario, the flight path parameters of the aerial photography device are determined according to the test image sequence of the test accident acquired by the aerial photography device, including: Establishing a dual-objective optimization model of the test image sequence according to the image overlap and path length between adjacent frame images in the test image sequence; By solving the dual-objective optimization model, the flight path parameters of the aerial photography equipment are obtained.

[0007] In some technical solutions of the present application, the dual-objective optimization model of the test image sequence is established according to the image overlap and the path length between adjacent frame images in the test image sequence, including: Establishing an overlap evaluation sub-model according to the original overlap ratio, image quality factor and terrain factor between adjacent frame images in the test image sequence; constructing a path efficiency sub-model according to the distance between adjacent frame images in the test image sequence and the flight difficulty coefficient; A dual-objective optimization model of the test image sequence is established according to the overlap evaluation sub-model and the path efficiency sub-model.

[0008] In some technical solutions of the present application, the flight path parameters of the aerial photography device are obtained by solving the dual-objective optimization model, including: generating a candidate path according to the image overlap restriction requirement and flight safety requirement; Generating a quality function for pheromone update according to the restriction requirements on the image overlap and the requirements on the path efficiency; The flight path parameters of the aerial photography device are determined according to the ant colony algorithm based on the quality function and the candidate path.

[0009] In some technical solutions of the present application, the first image sequence is obtained by: Inputting original scene data of the first accident scene into a feature parameter mapping model to obtain first shooting parameters of the first accident scene; During the process in which the aerial photography device flies according to the flight path parameters, the first frame of the first image in the first image sequence is captured according to the first shooting parameters; Optimizing the first shooting parameter by using a gradient descent method to obtain an optimized second shooting parameter; During the flight of the aerial photography device according to the flight path parameters, the second frame of the first image in the first image sequence is captured according to the second shooting parameters until the flight is completed.

[0010] In some technical solutions of the present application, the above method further includes: Calculate quality indicators during aerial photography; If the quality indicator meets the preset quality requirement, the flight path parameter and / or shooting parameter are adjusted.

[0011] In some technical solutions of the present application, the above method further includes: Determining the accident type of the first accident according to the accident characteristics of the first accident scene; If there is a second accident of the same accident type as the first accident, using the flight parameter set of the first accident to acquire a second image sequence of the second accident; The first accident scene obtained by reconstructing and simulating the second image sequence is displayed.

[0012] In a second aspect, an embodiment of the present application provides a device for accident restoration, the device comprising: A test module, used to determine the flight path parameters of the aerial photography device according to a test image sequence of a test accident acquired by the aerial photography device in a test scenario; An acquisition module is used to acquire a first image sequence taken by the aerial photography device in an actual scenario while the aerial photography device is flying according to the flight path parameters; wherein the first image sequence includes a plurality of first image frames, and a subsequent first image frame is obtained by optimizing the shooting parameters of a previous first image frame; A display module is used to display a first accident scene obtained by reconstructing and simulating the first image sequence.

[0013] In a third aspect, an embodiment of the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned accident restoration method when executing the computer program.

[0014] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above-mentioned accident restoration method are executed.

[0015] The technical solution provided by the embodiments of the present application may have the following beneficial effects: The method of the present application includes, in a test scenario, determining the flight path parameters of the aerial photography device according to a test image sequence of a test accident acquired by the aerial photography device; in an actual scenario, acquiring a first image sequence shot by the aerial photography device while the aerial photography device is flying according to the flight path parameters; wherein the first image sequence includes a plurality of frames of first images, and a subsequent frame of the first image is shot after optimizing the shooting parameters of a previous frame of the first image; and displaying a first accident scene reconstructed and simulated according to the first image sequence.

[0016] Through efficient data collection by drones and precise analysis by computer vision technology, this application can efficiently complete three-dimensional reconstruction and dynamic analysis of the accident scene, providing a scientific basis for the determination of accident responsibility.

[0017] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application 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 paying creative work.

[0019] Figure 1 A schematic diagram of a process of an accident restoration method provided in an embodiment of the present application is shown; Figure 2 A schematic diagram of the overall system structure of an accident restoration provided by an embodiment of the present application is shown; Figure 3 A schematic diagram of an accident restoration device provided in an embodiment of the present application is shown; Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0020] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It should be understood that the drawings in the present application only serve the purpose of explanation and description and are not used to limit the scope of protection of the present application. In addition, it should be understood that the schematic drawings are not drawn in real proportion. The flowchart used in this application shows the operations implemented according to some embodiments of the present application. It should be understood that the operations of the flowchart can be implemented out of sequence, and the steps without logical context can be reversed in order or implemented simultaneously. In addition, those skilled in the art can add one or more other operations to the flowchart under the guidance of the content of the present application, or remove one or more operations from the flowchart.

[0021] In addition, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0022] It should be noted that the term "comprising" will be used in the embodiments of the present application to indicate the existence of the features declared thereafter, but does not exclude the addition of other features.

[0023] The most important task of accident reconstruction is to use the remaining driving marks on the road at the accident scene, the collision marks on the vehicle body and the stopped position of the vehicle to estimate the vehicle's speed, acceleration, heading angle during the collision and the vehicle's driving trajectory after the collision, so as to infer the cause of the traffic accident, determine the responsibility for the accident and take corresponding measures.

[0024] Traffic accident scene investigation and accident analysis are important links in traffic management. Traditional accident scene investigation mainly relies on manual measurement and photographic evidence collection. This method has the following problems: Limitations of manual measurement: Measurement accuracy is limited by manual operation; the measurement process takes a long time; it is difficult to obtain complete spatial information; and it is easy to miss key details.

[0025] Deficiencies in accident reconstruction: lack of accurate three-dimensional spatial information; inability to conduct precise dynamic analysis; accident process reconstruction is not intuitive enough; analysis results are not convincing enough.

[0026] Data storage and sharing issues: Field data is difficult to preserve completely; post-analysis methods are limited; multi-department collaboration is inefficient; and archived data is not systematic enough.

[0027] With the development of technology, drone aerial photography technology, 3D reconstruction technology and vehicle dynamics simulation technology are becoming more and more mature, providing a possibility to solve the above problems. However, there is currently a lack of complete solutions that organically combine these technologies on the market.

[0028] Based on this, the embodiments of the present application provide a method, device, electronic device and storage medium for accident restoration, which are described below through embodiments.

[0029] Figure 1 A schematic flow chart of an accident restoration method provided in an embodiment of the present application is shown, wherein the method includes steps S101-S103; specifically: S101, in a test scenario, determining a flight path parameter of the aerial photography device according to a test image sequence of a test accident acquired by the aerial photography device; S102, in an actual scenario, while the aerial photography device is flying according to the flight path parameters, obtaining a first image sequence shot by the aerial photography device; wherein the first image sequence includes a plurality of first image frames, and a subsequent first image frame is shot after optimizing the shooting parameters of a previous first image frame; S103: Display a first accident scene obtained by reconstructing and simulating the first image sequence.

[0030] Through efficient data collection by drones and precise analysis by computer vision technology, this application can efficiently complete three-dimensional reconstruction and dynamic analysis of the accident scene, providing a scientific basis for the determination of accident responsibility.

[0031] Some embodiments of the present application are described in detail below. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0032] The embodiment of the present application provides a solution for accident recovery, and its system framework is as follows: Figure 2 As shown, it includes data acquisition layer, data processing layer, simulation analysis layer and result display layer. Among them, data acquisition layer: using drones to collect images of the accident scene; data processing layer: using modeling software to reconstruct the three-dimensional scene; simulation analysis layer: dynamics simulation in dynamics and kinematics simulation software; result display layer: using coding software to perform visual restoration. Specifically, use drones to shoot the accident scene according to the preset route, import the collected images into the modeling software, and generate an accurate three-dimensional model of the accident scene through steps such as feature extraction, camera orientation and dense reconstruction; then, import the three-dimensional model into the dynamics and kinematics simulation software, set vehicle parameters and environmental conditions, and perform vehicle collision dynamics simulation; finally, use coding software to process the simulation data to achieve three-dimensional scene rendering and visual restoration of vehicle motion trajectory.

[0033] When collecting data, in order to collect more accurate image data and improve shooting efficiency, the present application embodiment needs to further limit the flight path parameters and shooting parameters of the aerial photography equipment (drone). Specifically, in order to ensure shooting efficiency, the present application embodiment determines the shortest flight path parameters for the drone, and optimizes the shooting parameters in real time during the shooting process.

[0034] Specifically, the process of determining the flight path parameters of the shortest path is completed in a test scenario. That is, the embodiment of the present application determines the flight path parameters of the drone by shooting a test image sequence of a test accident with the drone and then analyzing the test image sequence.

[0035] When analyzing the test image sequence, the embodiment of the present application mainly considers the image overlap and path length between adjacent frame images. A dual-objective optimization model of the test image sequence is established based on the image overlap and path length between adjacent frame images. The dual objectives here are the minimum image overlap and the shortest path length. By solving the dual-objective optimization model, the flight path parameters of the aerial photography device are obtained.

[0036] When constructing a dual-objective optimization model, since it is a dual-objective optimization model, for the dual objectives, the embodiment of the present application needs to first construct two sub-models of the objectives: an overlap evaluation sub-model and a path efficiency sub-model. Specifically, based on the original overlap rate, image quality factor and terrain factor between adjacent frame images in the test image sequence, an overlap evaluation sub-model is established; based on the distance between adjacent frame images in the test image sequence and the flight difficulty coefficient, a path efficiency sub-model is constructed; based on the overlap evaluation sub-model and the path efficiency sub-model, a dual-objective optimization model of the test image sequence is established.

[0037] After the dual-objective optimization model is obtained, the flight path parameters of the UAV can be obtained by solving the dual-objective optimization model. Solving the dual-objective optimization model: generating a candidate path according to the restriction requirements on the image overlap and the flight safety requirements; generating a quality function for pheromone update according to the restriction requirements on the image overlap and the requirements on the path efficiency; determining the flight path parameters of the aerial photography device according to the ant colony algorithm based on the quality function and the candidate path.

[0038] In the specific implementation, a high-precision image sequence (test image sequence) of the test accident scene is obtained. The flight path planning of the drone is based on the following optimization objectives:

[0039] in represents the sequence of aerial photography path points, Indicates the overlap between adjacent aerial images; : Indicates the lack of overlap, the smaller the better; represents the total length of the path, is the weight coefficient used to balance overlap and path length. This optimization goal has two aspects: Minimize the lack of overlap: Make sure to get enough image overlap.

[0040] Minimize the flight path: Ensure flight efficiency.

[0041] By adjusting The value can balance these two goals: Smaller value: emphasis on image overlap; Larger value: Emphasis on path length optimization. By solving this optimization problem, an aerial photography path that meets the coverage requirements and has a shorter flight distance can be obtained.

[0042] First, a dual-objective optimization model based on overlap and path length was constructed. This model innovatively adopted a hierarchical structure design, consisting of an overlap evaluation sub-model and a path efficiency sub-model. The quantitative evaluation of the two objectives was achieved by introducing a fuzzy comprehensive evaluation index that considers image quality and terrain factors and an improved path cost function. At the same time, a scene adaptive weight adjustment mechanism was developed to dynamically balance the weight coefficients of the two sub-models according to the scene complexity, terrain undulations and other characteristics, so that the optimization results are more in line with actual application requirements. While ensuring the quality of aerial photography data, this model significantly improves the efficiency of path planning, and can reduce the flight distance by 20-30% compared with traditional methods.

[0043] Among them, the overlap evaluation sub-model adopts the following formula:

[0044] in, For adjacent routes and The comprehensive overlap evaluation value between is the original overlap ratio, is the image quality factor, is the terrain factor, , , is the dynamic weight coefficient.

[0045] The path efficiency submodel is calculated by the following formula:

[0046] in is the distance between adjacent waypoints, The flight difficulty coefficient is considered to take into account the altitude change and the turning angle. The final optimization objective function is:

[0047] This model significantly improves the efficiency of path planning while ensuring the quality of aerial photography data, and can reduce the flight distance by 20-30% compared to traditional methods.

[0048] The overlap target is The term is used to represent the image overlap between adjacent routes, and the path length target is used to represent the overall flight distance through the term L(P). The weight coefficient λ is used to balance these two goals. Through experiments on 30 different types of test accident sites, different values ​​of λ from 0.1 to 0.9 were tested. It was found that when λ = 0.3, a shorter flight path can be obtained while ensuring an 80% heading overlap rate and a 70% lateral overlap rate.

[0049] Specifically, the improved ant colony algorithm is used to solve this optimization problem, which mainly includes designing new pheromone update rules, developing a safety constraint heuristic information calculation method, and realizing adaptive adjustment of algorithm parameters. The core innovation lies in the design of new pheromone update rules:

[0050]

[0051] in is the pheromone volatility coefficient, is the overlap quality function, > is the weight coefficient. At the same time, the safety constraint heuristic information is introduced:

[0052]

[0053] in is the obstacle distance, is the height change, , is the adjustment coefficient.

[0054] The improved algorithm transforms the aerial photography overlap requirement into a quality function for pheromone updates, incorporates flight safety constraints into heuristic information calculations, and adaptively adjusts key parameters through an iterative process. Through these improvements, the algorithm reduces the calculation time by 35% and the path length by an average of 25% while ensuring the quality of route planning, significantly improving practicality while ensuring the integrity and reliability of the collected data.

[0055] The specific solution process is as follows: Optimizing aerial photography path planning needs to meet the following two main goals: Overlap requirement: Heading overlap rate ≥ 80%, lateral overlap rate ≥ 70%, to ensure the spatial coverage integrity between aerial images. Minimize path length: Minimize the total path length of the drone flight to save flight time and power.

[0056] The optimization objective function is:

[0057] : Path length, indicating flight efficiency; : Image overlap optimization target; : Weight coefficient used to balance the two objectives (usually adjusted in the range of 0.1 to 0.9).

[0058] Algorithm initialization: initialization parameters, pheromone matrix : The initial pheromone value between all waypoints is set to a small positive number (such as 0.1), indicating that the path has not been explored.

[0059] Heuristic Information : represents the heuristic probability of choosing a certain path, defined as:

[0060] in: :Waypoint and The horizontal distance between :Waypoint and The horizontal distance between : Adjustment coefficient, used to balance the impact of horizontal distance and height difference.

[0061] Parameter setting: Number of ants :Usually 50%-100% of the total number of waypoints is taken to ensure the comprehensiveness of the path exploration. : The general value range is [0.1, 0.5], which is used to control the dissipation of pheromones over time. Pheromone importance factor and heuristic information importance factor : The initial values ​​are set to 1 and 2, indicating that heuristic information is used first. Maximum number of iterations : For example, 100 times.

[0062] Path construction: Probabilistic path selection formula: When constructing a path, each ant starts from the current node Select next node The probability of is determined by the following formula:

[0063] : The pheromone value of the current path; : Heuristic information of the path; : The set of nodes that the ants have not visited yet; : Pheromone importance factor; : Heuristic information importance factor.

[0064] Path constraints: Overlap restriction: During the route construction process, the selected route must meet the requirements of heading and lateral overlap (≥80%, ≥70%); Flight safety: Safety constraints (e.g. obstacle distance, flight altitude limit) are introduced into the heuristic information to ensure that the path does not enter the dangerous area.

[0065] Path evaluation: For each ant's generated path , calculate the following indicators:

[0066] in For path segment distance.

[0067] Overlap optimization goal:

[0068] Indicates The actual overlap of the images, Indicates the target overlap (such as 80%).

[0069] Comprehensive fitness value:

[0070] Pheromone Update: Local pheromone updates: When each ant constructs a path, the pheromone of the path is updated in time:

[0071] in: , when the path When selected, increases pheromone strength.

[0072] Global pheromone updates: After all ants have completed the path construction, the global optimal path is selected , and strengthen and update it:

[0073] : Pheromone reinforcement factor, usually positively correlated with path fitness value.

[0074] Parameter adaptive adjustment: In order to improve the stability and adaptability of the algorithm, the following parameters are dynamically adjusted during the iteration process: Number of ants: As the number of iterations increases, the number of ants is gradually reduced to speed up convergence.

[0075] Pheromone volatilization rate: Take a lower value (such as 0.1) in the early stage to enhance the exploration ability, and gradually increase it in the later stage to prevent excessive search.

[0076] Weight coefficient λ: Dynamically adjusted according to changes in path length and overlap to adapt to different scenario requirements.

[0077] Convergence conditions: The algorithm stops when one of the following conditions is met: Reached the maximum number of iterations ; The global optimal path fitness value of the most recent multiple iterations Converged (change is less than threshold).

[0078] The specific flight path parameter settings are as follows: flight altitude: 30-50 meters; heading overlap: 80%; lateral overlap: 70%; ground resolution: 2 cm / pixel.

[0079] After obtaining the flight path parameters, the UAV is controlled to fly according to the flight path parameters at the first accident scene in the actual scene. The first accident scene is photographed during the flight to obtain a first image sequence of the first accident scene.

[0080] For the first frame of the first image sequence, the drone needs to shoot according to the preset first shooting parameters. The first shooting parameters here are determined by the original scene data of the first accident scene: the original scene data of the first accident scene is input into the feature parameter mapping model to obtain the first shooting parameters of the first accident scene. The original scene data here includes elevation data, image data (shot by other methods), etc., and then terrain feature extraction (calculation of elevation standard deviation, slope change rate, terrain undulation index), obstacle distribution analysis (spatial clustering identification and density calculation), and lighting condition evaluation (brightness analysis and environmental factor evaluation) are performed respectively. After processing, a standardized scene feature vector is output, which contains quantitative indicators such as terrain complexity index, obstacle distribution characteristics and lighting condition score, and the scene features are converted into the first shooting parameters through the feature-parameter mapping model.

[0081] After the first frame of the first image is captured, the first shooting parameters corresponding to the first frame of the first image are optimized to obtain the shooting parameters of the second frame of the first image. Then, the shooting parameters of the second frame of the first image are optimized to obtain the shooting parameters of the third frame of the first image. The first image sequence can be obtained by controlling the drone to shoot based on the shooting parameters.

[0082] For the optimization of shooting parameters, the gradient descent method is used, and safety constraints and quality requirements need to be considered. Specifically, in the path optimization of accident reconstruction, the goal is to meet the image overlap requirements: the heading overlap rate is ≥ 80%, and the lateral overlap rate is ≥ 70%. Shortest path length: reduce the flight path length as much as possible. Other requirements: such as flight safety (avoid obstacles, maintain a reasonable flight altitude).

[0083] These objectives can be expressed by the following optimization function:

[0084] in: : Path length, the distance of the total path segment. : Image overlap optimization objective. : Constraint penalty term, used to incorporate the part that violates the constraint into the objective function. λ: Weight coefficient, used to balance the path length and overlap.

[0085] Gradient descent optimization steps: (1) Constructing the objective function: By introducing the constraint requirements into the optimization goal, the following objective function is constructed:

[0086] in: :Waypoint and flight distance. : The first The actual overlap of the two images. : Target overlap (e.g. 80% in the heading direction and 70% in the sideways direction). : A penalty term used to represent constraint violation.

[0087] (2) Convert the constraints into penalty terms: To ensure that the path meets the constraints, the constraints can be embedded in the objective function in the form of a penalty function. For example: image overlap constraint: If the heading or lateral overlap of the image does not meet the requirements (such as 80%, 70%), a penalty function is introduced:

[0088] When the actual overlap Reaching the goal When , the penalty is 0; If the target value is not reached, the penalty value increases as the gap increases.

[0089] Flight altitude constraint: If the aerial photography flight altitude Beyond the safety range (for example, below 100 meters or above 500 meters), a penalty function is introduced:

[0090] : Minimum safe altitude. : Maximum safe altitude.

[0091] Obstacle distance constraint: If the flight path is close to an obstacle, the penalty function can be designed as the inverse of the obstacle distance:

[0092] : Current waypoint The distance to the nearest obstacle.

[0093] (3) Gradient calculation: According to the optimization objective function , calculate the gradient of each variable (such as the coordinates of the waypoint, the flight altitude, etc.). Assume that the variables are (indicates the location of the waypoint, flight altitude, etc.), the gradient is:

[0094] For example: Path length part:

[0095] Overlapping part:

[0096] Penalty section: .

[0097] (4) Gradient update: In each iteration, the waypoint positions or flight parameters are updated using gradient descent:

[0098] : Learning rate, controls the update step size.

[0099] (5) Projection to the feasible domain To ensure that the updated variables satisfy the constraints, a projection method can be used. For example, for the flight height constraint ≤ ≤ , after updating Projecting back to the feasible range:

[0100] Convergence judgment: When the objective function changes less than the threshold or reaches the maximum number of iterations, stop the optimization.

[0101] After the first image sequence is obtained, three-dimensional scene reconstruction is performed.

[0102] After the aerial data is collected, the modeling software is used to reconstruct the 3D scene. The reconstruction process is based on Structure from Motion (SfM) technology, the core of which is to solve the following optimization problems:

[0103] in, Indicates The projection matrix of the camera, Indicates The coordinates of the three-dimensional points, represents the observed two-dimensional image coordinates, Represents a visibility indicator; , calculate the square of the pixel distance between the 3D point projection and the actual observation point. This optimization process is implemented through bundle adjustment to obtain accurate camera pose and scene 3D structure.

[0104] The reconstruction process includes the following main steps: feature extraction and matching, camera pose estimation, sparse point cloud reconstruction, dense point cloud generation, triangular mesh construction, and texture mapping.

[0105] The accuracy of scene reconstruction is evaluated using the reprojection error:

[0106] in, is the actual observation point, is the reprojection point, n is the number of points involved in the evaluation, Used to calculate the average error.

[0107] Vehicle dynamics simulation: After completing the 3D scene reconstruction, the vehicle dynamics simulation is performed using dynamics and kinematics simulation software. The vehicle motion model is based on multi-body dynamics theory, and its motion equation can be expressed as:

[0108] in, is the mass matrix, is the damping matrix, is the stiffness matrix, are generalized coordinates, is the external force. The collision process adopts the elastic collision model, and the collision force can be expressed as:

[0109] in, is the contact stiffness, is the damping coefficient, is the deformation amount.

[0110] The main parameter settings during the simulation process include: Basic vehicle parameters: Mass distribution and moment of inertia, center of gravity location, track width and wheelbase; Suspension characteristics: Spring rate: , Damping coefficient: , anti-roll bar characteristics; The tire model uses the PAC2002 model:

[0111] in, is the characteristic coefficient, is the side slip angle.

[0112] Visual analysis and result display: Finally, use coding software to visualize the data and display the results. For the smoothing of the vehicle trajectory, the Savitzky-Golay filter is used:

[0113] in, is the smoothed data point, is the filter coefficient, The half-width of the window.

[0114] The velocity and acceleration are calculated using the central difference method:

[0115]

[0116] in: is location data, It's speed. is the acceleration, is the sampling time interval, and the central difference method is used to improve the calculation accuracy.

[0117] Trajectory Reproduction: def plot_trajectory(data): fig = plt.figure(figsize=(12, 8)) ax = fig.add_subplot(111, projection='3d') ax.plot3D(data['x'], data['y'], data['z']) Kinetic parameter analysis: def analyze_dynamics(data): # Calculate kinematic parameters velocity = np.gradient(data['position'], data['time']) acceleration = np.gradient(velocity, data['time']) # Calculate kinetic parameters kinetic_energy = 0.5 * data['mass'] * np.sum(velocity**2, axis=1) return velocity, acceleration, kinetic_energy Through the above scheme, this application realizes a complete workflow from on-site data collection to final visual analysis.

[0118] In an optional implementation, the embodiment of the present application continuously receives image quality data and current flight parameters during flight, calculates real-time quality indicators through multi-dimensional quality evaluation, and automatically calculates parameter adjustment amounts based on the quality loss type when the quality indicator is detected to be lower than a threshold. The module also processes historical adjustment data to optimize the adjustment strategy, and achieves dynamic optimization of flight parameters through this closed-loop feedback method.

[0119] Specifically: Multi-dimensional quality assessment, evaluating the quality indicators of multiple dimensions of aerial images: Overlap :Heading ≥ 80%, sideways ≥ 70%. Calculate the difference:

[0120] Clarity: Based on blur detection. Exposure: Detect overexposure / underexposure. Flight deviation: Evaluate the deviation between the actual and target waypoints.

[0121] Detection quality is below the threshold: If a certain indicator is below the preset threshold (such as overlap <80%, insufficient clarity, etc.), the adjustment process will be entered.

[0122] Dynamic parameter adjustment: Calculate the adjustment amount based on the type of quality problem: Insufficient overlap: Adjust the route spacing :

[0123] Insufficient clarity: Increase the shooting interval or adjust the focus.

[0124] Exposure issues: Dynamically adjust exposure compensation :

[0125] Flight deviation: Correcting waypoint coordinates :

[0126] Adjust parameters in real time: Update flight path, shooting time interval or camera parameters to ensure the next frame image quality meets the requirements.

[0127] In an optional implementation, after obtaining the flight parameter set of the first accident scene, the embodiment of the present application saves it to the knowledge base for subsequent use. The accident type of the first accident is also saved together with the flight parameter set.

[0128] The accident type of the first accident is mainly based on the terrain characteristics of the accident site (such as flat areas, mountains, and densely populated urban areas), environmental conditions (such as light intensity and weather conditions), and mission requirements (such as accident scale and complexity).

[0129] When there is a second accident of the same accident type as the first accident, the flight parameter set of the first accident can be directly adopted. Under the accident type, due to the similarity of characteristics and mission requirements, the optimized flight parameters (such as flight altitude, image overlap, flight speed, etc.) can be reused in similar scenarios. For example, the flight altitude and overlap standards in flat areas usually do not need to be adjusted, while the height and density need to be uniformly increased in complex terrain. By storing the optimal parameter configurations for these scene types in the knowledge base, the system can quickly call the parameter set suitable for the current task, reduce the computational burden, and ensure the efficiency and accuracy of the aerial photography mission.

[0130] Figure 3 A schematic diagram of the structure of an accident restoration device provided in an embodiment of the present application is shown, and the device includes: A test module, used to determine the flight path parameters of the aerial photography device according to a test image sequence of a test accident acquired by the aerial photography device in a test scenario; An acquisition module is used to acquire a first image sequence taken by the aerial photography device in an actual scenario while the aerial photography device is flying according to the flight path parameters; wherein the first image sequence includes a plurality of first image frames, and a subsequent first image frame is obtained by optimizing the shooting parameters of a previous first image frame; A display module is used to display a first accident scene obtained by reconstructing and simulating the first image sequence.

[0131] In the test scenario, determining the flight path parameters of the aerial photography device according to the test image sequence of the test accident acquired by the aerial photography device includes: Establishing a dual-objective optimization model of the test image sequence according to the image overlap and path length between adjacent frame images in the test image sequence; By solving the dual-objective optimization model, the flight path parameters of the aerial photography equipment are obtained.

[0132] The step of establishing a dual-objective optimization model of the test image sequence according to the image overlap and the path length between adjacent frame images in the test image sequence comprises: Establishing an overlap evaluation sub-model according to the original overlap ratio, image quality factor and terrain factor between adjacent frame images in the test image sequence; constructing a path efficiency sub-model according to the distance between adjacent frame images in the test image sequence and the flight difficulty coefficient; A dual-objective optimization model of the test image sequence is established according to the overlap evaluation sub-model and the path efficiency sub-model.

[0133] The step of solving the dual-objective optimization model to obtain the flight path parameters of the aerial photography device includes: generating a candidate path according to the image overlap restriction requirement and flight safety requirement; Generating a quality function for pheromone update according to the restriction requirements on the image overlap and the requirements on the path efficiency; The flight path parameters of the aerial photography device are determined according to the ant colony algorithm based on the quality function and the candidate path.

[0134] The first image sequence is obtained by: Inputting original scene data of the first accident scene into a feature parameter mapping model to obtain first shooting parameters of the first accident scene; During the process in which the aerial photography device flies according to the flight path parameters, the first frame of the first image in the first image sequence is captured according to the first shooting parameters; Optimizing the first shooting parameter by using a gradient descent method to obtain an optimized second shooting parameter; During the flight of the aerial photography device according to the flight path parameters, the second frame of the first image in the first image sequence is captured according to the second shooting parameters until the flight is completed.

[0135] Adjustment module, used to calculate the quality indicators during the aerial photography process; If the quality indicator meets the preset quality requirement, the flight path parameter and / or shooting parameter are adjusted.

[0136] a determination module, configured to determine the accident type of the first accident according to accident characteristics of the first accident scene; If there is a second accident of the same accident type as the first accident, using the flight parameter set of the first accident to acquire a second image sequence of the second accident; The first accident scene obtained by reconstructing and simulating the second image sequence is displayed.

[0137] like Figure 4 As shown, an embodiment of the present application provides an electronic device for executing the accident restoration method in the present application, the device includes a memory, a processor, a bus, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the accident restoration method when executing the computer program.

[0138] Specifically, the above-mentioned memory and processor may be general-purpose memory and processor, which are not specifically limited here. When the processor runs the computer program stored in the memory, the above-mentioned accident restoration method can be executed.

[0139] Corresponding to the accident restoration method in the present application, an embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is run by a processor, the steps of the above-mentioned accident restoration method are executed.

[0140] Specifically, the storage medium can be a general storage medium, such as a mobile disk, a hard disk, etc. When the computer program on the storage medium is run, the above-mentioned accident recovery method can be executed.

[0141] In the embodiments provided in the present application, it should be understood that the disclosed systems and methods can be implemented in other ways. The system embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, indirect coupling or communication connection of systems or units, which can be electrical, mechanical or other forms.

[0142] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0143] In addition, each functional unit in the embodiments provided in the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0144] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program codes.

[0145] It should be noted that similar numbers and letters represent similar items in the following figures. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In addition, the terms "first", "second", "third", etc. are only used to distinguish the description and are not to be understood as indicating or implying relative importance.

[0146] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present application, which are used to illustrate the technical solutions of the present application, rather than to limit them. The protection scope of the present application is not limited thereto. Although the present application is described in detail with reference to the above-mentioned embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-mentioned embodiments within the technical scope disclosed in the present application, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application. They should all be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A method for accident restoration, characterized in that: The method comprises: In a test scenario, determining a flight path parameter of the aerial photography device according to a test image sequence of a test accident acquired by the aerial photography device; In an actual scenario, when the aerial photography device flies according to the flight path parameters, a first image sequence shot by the aerial photography device is obtained; wherein the first image sequence includes a plurality of first image frames, and a subsequent first image frame is shot after optimizing the shooting parameters of a previous first image frame; The first accident scene obtained by reconstructing and simulating the first image sequence is displayed.

2. The method according to claim 1, characterized in that In the test scenario, determining the flight path parameters of the aerial photography device according to the test image sequence of the test accident acquired by the aerial photography device includes: Establishing a dual-objective optimization model of the test image sequence according to the image overlap and path length between adjacent frame images in the test image sequence; By solving the dual-objective optimization model, the flight path parameters of the aerial photography equipment are obtained.

3. The method according to claim 2, characterized in that The step of establishing a dual-objective optimization model of the test image sequence according to the image overlap and the path length between adjacent frame images in the test image sequence comprises: Establishing an overlap evaluation sub-model according to the original overlap ratio, image quality factor and terrain factor between adjacent frame images in the test image sequence; constructing a path efficiency sub-model according to the distance between adjacent frame images in the test image sequence and the flight difficulty coefficient; A dual-objective optimization model of the test image sequence is established according to the overlap evaluation sub-model and the path efficiency sub-model.

4. The method according to claim 2, characterized in that: The step of solving the dual-objective optimization model to obtain the flight path parameters of the aerial photography device includes: generating a candidate path according to the image overlap restriction requirement and flight safety requirement; Generating a quality function for pheromone update according to the restriction requirements on the image overlap and the requirements on the path efficiency; The flight path parameters of the aerial photography device are determined according to the ant colony algorithm based on the quality function and the candidate path.

5. The method according to claim 1, characterized in that The first image sequence is obtained by: Inputting original scene data of the first accident scene into a feature parameter mapping model to obtain first shooting parameters of the first accident scene; During the process in which the aerial photography device flies according to the flight path parameters, the first frame in the first image sequence is captured according to the first shooting parameters; Optimizing the first shooting parameter by using a gradient descent method to obtain an optimized second shooting parameter; During the flight of the aerial photography device according to the flight path parameters, the second frame of the first image in the first image sequence is captured according to the second shooting parameters until the flight is completed.

6. The method according to claim 1, characterized in that The method further comprises: Calculate quality indicators during aerial photography; If the quality indicator meets the preset quality requirement, the flight path parameter and / or shooting parameter are adjusted.

7. The method according to claim 1, characterized in that The method further comprises: Determining the accident type of the first accident according to the accident characteristics of the first accident scene; If there is a second accident of the same accident type as the first accident, using the flight parameter set of the first accident to acquire a second image sequence of the second accident; The first accident scene obtained by reconstructing and simulating the second image sequence is displayed.

8. An accident recovery device, characterized in that: The device comprises: A test module, used to determine the flight path parameters of the aerial photography device according to a test image sequence of a test accident acquired by the aerial photography device in a test scenario; An acquisition module is used to acquire a first image sequence taken by the aerial photography device in an actual scenario while the aerial photography device is flying according to the flight path parameters; wherein the first image sequence includes a plurality of first image frames, and a subsequent first image frame is obtained by optimizing the shooting parameters of a previous first image frame; A display module is used to display a first accident scene obtained by reconstructing and simulating the first image sequence.

9. An electronic device, characterized in that: include: A processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate via the bus, and when the machine-readable instructions are executed by the processor, the steps of the accident restoration method as described in any one of claims 1 to 7 are performed.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which, when executed by a processor, executes the steps of the accident restoration method as described in any one of claims 1 to 7.

Images