Method and device for drawing a vector diagram of an accident scene

Abstract

This disclosure provides a method and apparatus for drawing a vector map of an accident scene. The method includes: performing object recognition on a global top-view of a target accident scene to determine at least one object included in the target accident scene; symbolizing the at least one object in the global top-view to obtain at least one object symbol; determining at least one reference element based on the distribution of the at least one object in the global top-view; determining annotation information for at least one key accident point in the global top-view based on the at least one reference element; and obtaining a symbolized vector map of the scene with annotation information based on the annotation information of the at least one key accident point and the at least one object symbol. This disclosure enables the automatic drawing of a standard vector map of a target accident scene while automatically annotating the details of the accident scene.

Description

Technical Field

[0001] This disclosure relates to the field of intelligent transportation, and in particular to a method and apparatus for drawing vector diagrams of accident scenes. Background Technology

[0002] In traffic accident scene investigation, the industry still widely uses traditional methods such as camera photography and manual measurement to collect on-site data and draw accident scene maps. Under this traditional investigation model, key physical parameters such as distances and dimensions are obtained through manual point-by-point measurement. This not only involves a large workload and is time-consuming, but manual operation is also easily affected by factors such as the on-site environment and operator experience, resulting in inconsistent data accuracy. Although emerging technologies such as 3D lasers have been gradually introduced for point cloud acquisition, practical applications still face the technical bottleneck of "macroscopic and microscopic separation," making it difficult to clearly present subtle features such as vehicle scratches and collision points. This makes it difficult for investigators to accurately pick out key points in 3D space, often requiring tedious manual operations to complete measurements and correlations. Summary of the Invention

[0003] To address the aforementioned technical problems, this disclosure is proposed. Embodiments of this disclosure provide a method and apparatus for drawing a vector diagram of an accident scene.

[0004] According to one aspect of the present disclosure, a method for drawing a vector diagram of an accident scene is provided, comprising: Object identification is performed on a global top-down view of the target accident scene to determine at least one object included in the target accident scene; At least one object in the global top view is symbolized to obtain at least one object symbol; each object symbol corresponds to one object. Based on the distribution of the at least one object symbol in the global top view, at least one reference element is determined; Based on the at least one reference element, the annotation information of at least one accident key point in the global top view is determined; each accident key point corresponds to one of the objects. Based on the annotation information of the at least one key accident point and the at least one object symbol, a symbolic on-site vector map with annotation information is obtained.

[0005] Optionally, determining at least one reference feature based on the distribution of the at least one object symbol in the global top view includes: The distribution of the at least one object symbol in the global top view is determined based on the position of the at least one object symbol in the global top view; Based on the distribution of the at least one object symbol in the global top view, the at least one object symbol located at a preset position in the distribution is determined as the at least one reference element.

[0006] Optionally, it also includes: Acquire high-resolution images of the at least one object; each object corresponds to at least one high-resolution image; The step of determining the annotation information of at least one accident key point in the global top view based on the at least one reference feature includes: Based on the click command received in the high-definition image or by identifying the accident-related area in the high-definition image, determine the image coordinates of the accident key points in the object in the global top view; The target reference element is determined from the at least one reference element based on the image coordinates of the key points of the accident; Based on the positional relationship between the accident key point and the target reference element, and the positional relationship between the at least one accident key point, the annotation information of the accident key point is determined.

[0007] Optionally, determining the annotation information of the key accident points based on the positional relationship between the key accident points and the target reference elements includes: Determine the type of the target reference feature; the type of the reference feature includes reference points and / or reference lines; In response to the target reference element being the reference line, the annotation information is determined based on the vertical distance between the accident critical point and the reference line; and / or, In response to the target reference element being the reference point, the annotation information is determined based on the line segment length between the accident critical point and the reference point.

[0008] Optionally, determining the image coordinates of key accident points in the object in the global top view based on a click command received in the high-definition image or by identifying an accident-related region in the high-definition image includes: Based on the click command received in the high-definition image or by identifying key points of the accident in the high-definition image, determine the coordinates of key points in the high-definition image; Based on the correspondence between the high-definition image and the point cloud data, the coordinates of the key points are transformed to determine the target three-dimensional coordinates of the key accident points in the point cloud data. Based on the projection relationship between the point cloud data and the global top view, the image coordinates corresponding to the three-dimensional coordinates of the target are determined.

[0009] Optionally, obtaining a symbolized on-site vector map with annotation information based on the annotation information of the at least one key accident point and the at least one object symbol includes: Based on the positional relationship between the at least one key accident point, establish a key point association line segment, and display the attribute information between the key accident points on the key point association line segment; Establish marker line segments from the key accident points to the corresponding reference elements, and display the annotation information corresponding to the key accident points on the marker line segments to obtain the on-site vector map.

[0010] Optionally, the symbolization of at least one object in the global top view to obtain at least one object symbol includes: Identify the size of at least one object included in the global top view, as well as the category and / or shape of the object; Determine the initial object symbol corresponding to the object based on the object's category and / or shape; The size of the initial object symbol is adjusted according to the size of the object to obtain the object symbol of the at least one object in the global top view.

[0011] Optionally, after performing object identification on the global top-down view of the target accident scene to determine that the target accident scene includes at least one object, the method further includes: Based on the point cloud data corresponding to the at least one object, determine the position information and orientation of the at least one object; The step of adjusting the size of the initial object symbol according to the size of the object to obtain the object symbol of the at least one object in the global top view includes: According to the position information of the at least one object, insert the at least one initial object symbol into the global top view; The at least one initial object symbol is stretched according to the orientation of the at least one target object until the length of the initial object symbol matches the size of the target object, thus obtaining the object symbol.

[0012] Optionally, the step of performing object identification on a global top-down view of the target accident scene to determine that the target accident scene includes at least one object, including: The global top view is determined based on the point cloud data collected from the target accident site; Identify at least one region of interest in the global top view and determine at least one object included in the target accident scene.

[0013] According to another aspect of the present disclosure, an apparatus for drawing a vector diagram of an accident scene is provided, comprising: The object recognition module is used to perform object recognition on a global top view of the target accident scene and determine that the target accident scene includes at least one object. An object symbolization module is used to symbolize at least one object in the global top view to obtain at least one object symbol; each object symbol corresponds to one object. A benchmark determination module is used to determine at least one benchmark element based on the distribution of the at least one object symbol in the global top view; The annotation information module is used to determine the annotation information of at least one accident key point in the global top view based on the at least one reference element; each accident key point corresponds to one object; The vectorization module is used to obtain a symbolic on-site vector map with annotation information based on the annotation information of the at least one key accident point and the at least one object symbol.

[0014] Optionally, the reference determination module is specifically used to determine the distribution of the at least one object symbol in the global top view based on the position of the at least one object symbol in the global top view; and to determine the at least one object symbol located at a preset position in the distribution as the at least one reference element based on the distribution of the at least one object symbol in the global top view.

[0015] Optionally, the device further includes: An image acquisition module is used to acquire high-resolution images of the at least one object; each object corresponds to at least one high-resolution image; The annotation information module includes: The coordinate determination unit is used to determine the image coordinates of the key accident points in the object in the global top view based on the click command received in the high-definition image or the identification of the accident-related area in the high-definition image. An element filtering unit is used to determine a target reference element from the at least one reference element based on the image coordinates of the key points of the accident. A standard determination unit is used to determine the annotation information of the accident key point based on the positional relationship between the accident key point and the target reference element, and the positional relationship between the at least one accident key point.

[0016] Optionally, the standard determination unit is specifically used to determine the type of the target reference element; the type of the reference element includes a reference point and / or a reference line; in response to the target reference element being the reference line, the labeling information is determined based on the vertical distance between the accident key point and the reference line; and / or, in response to the target reference element being the reference point, the labeling information is determined based on the line segment length between the accident key point and the reference point.

[0017] Optionally, the coordinate determination unit is specifically configured to: determine the coordinates of key points in the high-definition image based on a click instruction received in the high-definition image or by identifying key accident points in the high-definition image; convert the coordinates of the key points according to the correspondence between the high-definition image and the point cloud data to determine the target three-dimensional coordinates corresponding to the key accident points in the point cloud data; and determine the image coordinates corresponding to the target three-dimensional coordinates according to the projection relationship between the point cloud data and the global top view.

[0018] Optionally, the vectorization module is specifically used to establish key point association line segments based on the positional relationship between the at least one key accident point, and display the attribute information between the key accident points on the key point association line segments; establish identification line segments from the key accident point to the corresponding reference element, and display the annotation information corresponding to the key accident point on the identification line segments, thereby obtaining the on-site vector map.

[0019] Optionally, the object symbolization module is specifically used to identify the size of at least one object included in the global top view, as well as the category and / or shape of the object; determine the initial object symbol corresponding to the object based on the category and / or shape of the object; and adjust the size of the initial object symbol according to the size of the object to obtain the object symbol of the at least one object in the global top view.

[0020] Optionally, the object symbolization module is further configured to determine the position information and orientation of the at least one object based on the point cloud data corresponding to the at least one object; When the object symbolization module adjusts the size of the initial object symbol according to the size of the object to obtain the object symbol of the at least one object in the global top view, it inserts the at least one initial object symbol into the global top view according to the position information of the at least one object; and performs stretching processing on the at least one initial object symbol according to the orientation of the at least one target object until the length of the initial object symbol matches the size of the target object to obtain the object symbol.

[0021] Optionally, the object recognition module is specifically used to determine the global top view based on the point cloud data collected from the target accident scene; identify at least one region of interest in the global top view; and determine at least one object included in the target accident scene.

[0022] According to another aspect of the present disclosure, an electronic device is provided, comprising: Memory, used to store computer program products; A processor is configured to execute a computer program product stored in the memory, and when the computer program product is executed, to implement the method for drawing an accident scene vector diagram as described in any of the above embodiments.

[0023] According to another aspect of the present disclosure, a computer-readable storage medium is provided, on which computer program instructions are stored, which, when executed by a processor, implement the method for drawing an accident scene vector diagram as described in any of the above embodiments.

[0024] According to another aspect of the present disclosure, a computer program product is provided, including computer program instructions that, when executed by a processor, implement the method for drawing an accident scene vector diagram as described in any of the above embodiments.

[0025] Based on the method and apparatus for drawing accident scene vector diagrams provided in the above embodiments of this disclosure, object recognition is performed on a global top view of a target accident scene to determine at least one object included in the target accident scene; objects at the accident scene are quickly and automatically acquired, providing a basis for subsequent object drawing; at least one object in the global top view is symbolized to obtain at least one object symbol; standardization of drawing is achieved through object symbolization; at least one reference element is determined based on the distribution of the at least one object in the global top view; by determining at least one reference element in the global top view, a reference object is provided for object annotation, and a unified standard annotation for at least one object can be achieved through the reference object; annotation information of at least one key accident point in the global top view is determined based on the at least one reference element; based on the annotation information of the at least one key accident point and the at least one object symbol, a symbolized scene vector diagram with annotation information is obtained. This disclosure embodiment determines the key points of the accident for key points related to the target accident (e.g., collision points), and marks the key points of the accident in combination with reference elements, thereby realizing automatic annotation of the target accident scene. This avoids drawing errors caused by manual drawing and annotation, and solves the problem that it is impossible to annotate the accident scene in detail by relying solely on point cloud data. Furthermore, by combining symbolic object symbols, it realizes the automatic drawing of the standard vector map of the target accident scene while automatically annotating the details of the accident scene.

[0026] The technical solutions of this disclosure will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0027] The accompanying drawings, which form part of this specification, illustrate embodiments of this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0028] This disclosure will become clearer with reference to the accompanying drawings and the following detailed description, wherein: Figure 1 This is a flowchart illustrating a method for drawing a vector map of an accident scene provided in an exemplary embodiment of this disclosure; Figure 2 This is a schematic diagram of a global top view in a method provided by an exemplary embodiment of this disclosure; Figure 3 This is a flowchart illustrating the process of determining an object symbol in an exemplary embodiment of this disclosure; Figure 4 This is a flowchart illustrating the process of determining an object symbol in a method provided in another exemplary embodiment of this disclosure; Figure 5 This is a flowchart illustrating the process of determining a reference element in an exemplary embodiment of this disclosure; Figure 6 This is a schematic flowchart illustrating the method for determining distance information provided in an exemplary embodiment of this disclosure; Figure 7 This is a schematic diagram showing the annotation information of at least one key point of an accident in the method provided in the optional example of Embodiment 1 of this disclosure; Figure 8 This is a flowchart illustrating the process of marking key points of an accident in an exemplary embodiment of the present disclosure; Figure 9 This is a schematic diagram of a vector graphic generated in a method provided in another exemplary embodiment of this disclosure; Figure 10 This is a schematic diagram of the structure of an apparatus for drawing vector diagrams of accident scenes provided in an exemplary embodiment of this disclosure; Figure 11 A block diagram of an electronic device according to an embodiment of the present disclosure is shown. Detailed Implementation

[0029] Hereinafter, exemplary embodiments according to the present disclosure will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of the present disclosure, and not all embodiments of the present disclosure, and it should be understood that the present disclosure is not limited to the exemplary embodiments described herein.

[0030] It should be noted that, unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps set forth in these embodiments do not limit the scope of this disclosure.

[0031] Those skilled in the art will understand that the terms "first," "second," etc., in the embodiments of this disclosure are only used to distinguish different steps, devices, or modules, and do not represent any specific technical meaning, nor do they indicate a necessary logical order between them.

[0032] It should also be understood that in the embodiments disclosed herein, "a plurality of" may refer to two or more, and "at least one" may refer to one, two or more.

[0033] It should also be understood that any component, data or structure mentioned in the embodiments of this disclosure can generally be understood as one or more unless expressly defined or given to the contrary in the context.

[0034] Furthermore, the term "and / or" in this disclosure is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this disclosure generally indicates that the preceding and following related objects have an "or" relationship. The data referred to in this disclosure can include unstructured data such as text, images, and videos, as well as structured data.

[0035] It should also be understood that the description of the various embodiments in this disclosure emphasizes the differences between the various embodiments, and the similarities or similarities can be referred to each other. For the sake of brevity, they will not be described in detail.

[0036] At the same time, it should be understood that, for ease of description, the dimensions of the various parts shown in the accompanying drawings are not drawn according to actual scale.

[0037] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit this disclosure or its application or use.

[0038] Techniques, methods, and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and equipment should be considered part of the specification.

[0039] It should be noted that similar labels and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be discussed further in subsequent figures.

[0040] The embodiments disclosed herein can be applied to electronic devices such as terminal devices, computer systems, and servers, and can operate together with a wide range of other general-purpose or special-purpose computing system environments or configurations. Examples of well-known terminal devices, computing systems, environments, and / or configurations suitable for use with electronic devices such as terminal devices, computer systems, and servers include, but are not limited to: personal computer systems, server computer systems, thin clients, thick clients, handheld or laptop devices, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputer systems, mainframe computer systems, and distributed cloud computing environments including any of the above systems, etc.

[0041] Electronic devices such as terminal devices, computer systems, and servers can be described in the general context of computer system executable instructions (such as program modules) executed by a computer system. Typically, program modules can include routines, programs, object programs, components, logic, data structures, etc., which perform specific tasks or implement specific abstract data types. Computer systems / servers can be implemented in distributed cloud computing environments, where tasks are executed by remote processing devices linked through communication networks. In distributed cloud computing environments, program modules can reside on local or remote computing system storage media, including storage devices.

[0042] Exemplary methods Figure 1 This is a flowchart illustrating a method for drawing a vector map of an accident scene according to an exemplary embodiment of this disclosure. This embodiment can be applied to electronic devices, such as... Figure 1 As shown, it includes the following steps: Step 102: Perform object identification on the global top view of the target accident scene to determine at least one object included in the target accident scene.

[0043] Optionally, this embodiment can scan the target accident scene using at least one sensor (e.g., LiDAR, depth camera, etc.) to obtain three-dimensional point cloud data (optionally, the point cloud data can be high-density point cloud data). Since the point cloud data has three-dimensional coordinate information, a global top-down view of the target accident scene can be obtained based on operations such as top-down projection. Optionally, the global top-down view can be a rasterized top-down view or a real-scene top-down view. This embodiment quickly acquires point cloud data using mobile devices, obtaining a complete and accurate three-dimensional digital base map of the scene, providing a clear, accurate, and objective data source for all subsequent operations.

[0044] Optionally, the target accident scene can be a traffic accident scene, a production safety accident scene (an accident that occurs during production, construction, or operation), a public safety accident scene (an accident that occurs in public places or public facilities), an accident scene caused by natural disasters (e.g., accidents caused by natural factors such as earthquakes, floods, and typhoons), or other accident scenes in the transportation sector (accidents can occur in all areas of transportation except for road traffic accidents, such as railway accident scenes and water traffic accident scenes).

[0045] In this embodiment, the global top view may include at least one object, and each of the at least one object may be processed separately. Optionally, at least one object may be processed sequentially, or at least one object may be processed simultaneously.

[0046] Optionally, the object can be any object in the target accident scene, such as vehicles, people, electric vehicles, road lines, signals, road edges, collision marks, etc. The global top-down view can be a top-down view of the scene obtained by image acquisition or point cloud acquisition of the target accident scene and then processing it. Existing technologies in the fields of computer vision and image processing can be used to identify at least one object in the global top-down view. For example, the object recognition method can include, but is not limited to, at least one of the following: recognition methods based on low-level features such as color, grayscale, and texture; recognition methods based on edge detection, contour extraction, and geometric transformation; recognition methods based on template matching and feature point matching; and object detection, semantic segmentation, and instance segmentation methods based on deep learning.

[0047] Step 104: Symbolize at least one object in the global top view to obtain at least one object symbol.

[0048] Each object symbol corresponds to one object.

[0049] Optionally, object symbolization can be achieved by matching multiple pre-set symbols with objects, thus supporting the standardization of drawing by symbolizing objects in the global top view.

[0050] In one embodiment, at least one object included in the global top view can be identified using deep learning methods or other object recognition methods (e.g., connected component analysis). Optionally, the global top view can be semantically segmented using a semantic segmentation model to obtain a category label + object mask corresponding to at least one object. The category of the object can be directly determined based on the category label, and the shape and size of the object can be determined based on the shape and size of the mask. Optionally, connected component analysis can be used to first binarize the global top view to obtain a binarized top view. Neighborhood connected component detection can be performed on the binarized top view to output a mask + bounding rectangle corresponding to each object. The size and shape of the target object can then be determined based on the mask and the bounding rectangle. This achieves symbolic management of target accident scene objects, avoids shape differences caused by manual drawing, and realizes unified management of object drawing through symbolic objects.

[0051] Step 106: Determine the annotation information of at least one accident key point in the global top view based on at least one reference element.

[0052] Each critical point of an accident corresponds to an object, and an object may correspond to one or more critical points of an accident, or an object may not correspond to any critical point of an accident.

[0053] Optionally, the key points of the accident are the key points in the object symbol (e.g., the collision part, etc.). For example, the key points of the accident can be determined by region recognition or user click. Based on the key points of the accident, the accident details involved in the target accident scene can be automatically marked, saving manpower and improving the accuracy of detail marking.

[0054] This embodiment uses a baseline element as a fixed reference point, associates the key accident points in the object symbol with the baseline element, and uses the distance information between the key accident points and the baseline element as the attribute information describing the key accident points. This enables all key accident points at the same accident scene to be described with reference elements, thus achieving standardized drawing and description of the accident scene.

[0055] Step 108: Determine at least one reference feature based on the distribution of at least one object in the global top view.

[0056] Optionally, the target accident scene typically has fixed scene markers. For example, traffic accident scenes usually include lane lines, edge lines, stop lines, guardrails, bridge piers, traffic lights, signposts, milestones, and other fixed scene markers. Optionally, the user can select one or more scene markers as reference elements by clicking. This embodiment can determine scene markers as reference elements, describe the distance between the object and the reference elements, realize the datafication and objectification of the location, and allow others to repeatedly measure and verify based on the same reference element, making the conclusions verifiable.

[0057] Step 110: Based on the annotation information of at least one key accident point and at least one object symbol, obtain a symbolized on-site vector map with annotation information.

[0058] In this embodiment, after determining the annotation information of the key accident points relative to the baseline elements, the key accident points are annotated in the global top view based on this annotation information, thereby realizing the attribute description of the key accident points in the global top view. When the target accident scene involves multiple objects and their corresponding key accident points, the accident scene can be described more accurately and intuitively. Furthermore, combined with the symbolized object symbols, the standard vector drawing of the accident scene is realized, so that the resulting vector map can intuitively and realistically restore the various objects in the accident scene and the relationships between them.

[0059] The method provided in the above embodiments of this disclosure performs object recognition on a global top-down view of a target accident scene to determine at least one object included in the target accident scene; achieves rapid and automatic acquisition of objects at the accident scene, providing a basis for subsequent object drawing; performs symbolization processing on at least one object in the global top-down view to obtain at least one object symbol; achieves standardization of drawing through object symbolization; determines at least one reference element based on the distribution of the at least one object in the global top-down view; provides a reference object for object annotation by determining at least one reference element in the global top-down view, and achieves unified standard annotation for at least one object through the reference object; determines the annotation information of at least one key accident point in the global top-down view based on the at least one reference element; and obtains a symbolized scene vector map with annotation information based on the annotation information of the at least one key accident point and the at least one object symbol. This disclosure embodiment determines the key points of the accident for key points related to the target accident (e.g., collision points), and marks the key points of the accident in combination with reference elements, thereby realizing automatic annotation of the target accident scene. This avoids drawing errors caused by manual drawing and annotation, and solves the problem that it is impossible to annotate the accident scene in detail by relying solely on point cloud data. Furthermore, by combining symbolic object symbols, it realizes the automatic drawing of the standard vector map of the target accident scene while automatically annotating the details of the accident scene.

[0060] This embodiment provides a method for rapid, accurate, and standardized accident scene investigation and mapping based on point cloud and image technology on a mobile terminal. It fundamentally transforms the traditional purely manual investigation mode, solving its three core problems of efficiency bottleneck, lack of accuracy, and insufficient standardization. It quickly completes the drawing of "accident scene map", freeing investigators from heavy manual labor and shortening the entire process time from "hours" to "minutes" from on-site investigation to the generation of standard scene map. It automatically generates dimension lines and values ​​that conform to national standards, and finally forms a 1:1 scene map based on point cloud, while significantly improving the accuracy and evidentiary value of the scene map.

[0061] In some alternative embodiments, based on the above embodiments, step 102 may include: A global top-down view is determined based on point cloud data collected from the target accident site.

[0062] Optionally, the point cloud data can be a colorless 3D point cloud or a high-density colored point cloud. When the point cloud data is a colorless 3D point cloud, it can be obtained by point cloud acquisition using LiDAR or a depth camera. However, when the point cloud data is a high-density colored point cloud, it needs to be combined with a real-world image. The real-world image can be obtained based on RGB camera acquisition. For example, a Simultaneous Localization and Mapping Device (SLAM device) that integrates multiple sensors (LiDAR + RGB camera, etc.) can be used to simultaneously acquire point cloud and color image to obtain a high-density colored point cloud.

[0063] Optionally, the 3D point cloud data can be mapped onto a 2D top-view plane to obtain a global top view.

[0064] The process of obtaining a global top view may include: fixing the Z-axis height, determining an XY plane as the top view reference, and mapping the point cloud data to the XY plane to obtain the global top view. Figure 2 This is a schematic diagram of a global top view in a method provided by an exemplary embodiment of this disclosure. For example... Figure 2 As shown, the target accident scene in this embodiment is a traffic accident scene, and the accident-related objects involved in the figure include cars, bicycles, lane lines, etc.

[0065] Identify at least one region of interest in the global top view and determine at least one object included in the target accident scene.

[0066] Optionally, common methods for identifying target objects (such as vehicles, lane lines, traffic lights, signs, pedestrians, collision points, brake marks, road edges, etc.) in images, which belong to the existing technology in the field of computer vision and image processing, may include, but are not limited to, at least one of the following: recognition methods based on color and texture features: By extracting color, grayscale, and texture features of target objects in an image and setting thresholds or feature templates, the target can be distinguished from the background. For example, object segmentation and localization can be achieved based on the white / yellow features of lane lines and the grayscale differences between the road surface and brake marks. Edge and contour detection-based recognition methods: Edge detection operators (such as Canny, Sobel, and Laplacian) are used to extract edges, contours, corners, and lines from the image. Morphological processing and contour filtering are then used to obtain the shape and position of the target object. This method is suitable for recognizing objects with clear boundaries, such as road edges, lane lines, and vehicle contours. Geometric feature recognition based on Hough transform: Hough line transform and Hough circle transform are used on edge images to detect geometric structures such as lines, arcs, and curves. This is commonly used for the recognition and fitting of regular geometric objects such as lane lines, road edges, and guardrails. Template matching-based recognition methods: Standard templates for target objects (such as traffic lights, traffic signs, and typical vehicle outlines) are pre-established. Similarity matching is performed between the target object and the template using a sliding window in the image, and the target location and category are determined based on the matching degree. Feature point detection and description-based recognition methods: Local feature points in the image are extracted using algorithms such as SIFT, SURF, ORB, and FAST to generate feature descriptors. Target object detection, localization, and tracking are achieved through feature matching. Semantic segmentation and instance segmentation-based recognition methods: Deep learning models (such as FCN, U-Net, Mask R-CNN, and DeepLab) are used to perform pixel-level classification of the image, labeling each pixel with a corresponding category (road, vehicle, traffic light, lane line, etc.) to achieve accurate segmentation, localization, and contour extraction of the target object. Recognition methods based on object detection networks: These methods employ deep learning object detection models (such as YOLO, Faster R-CNN, SSD, etc.) to directly predict the object category and bounding box in the image, enabling rapid detection, localization, and recognition of multiple object types. Recognition methods based on the fusion of depth information and 3D data: These methods combine RGB images with depth maps and point cloud data, utilizing depth values, 3D coordinates, height differences, and normal vectors to assist in recognition. This effectively eliminates occlusion, lighting, and perspective interference in 2D images, improving the robustness of object recognition in complex scenes.

[0067] Figure 3 This is a flowchart illustrating the process of determining an object symbol in an exemplary embodiment of this disclosure. Figure 3 As shown above, in the above Figure 1 Based on the illustrated embodiment, step 104 may include the following steps: Step 1041: Identify the size of at least one object included in the global top view, as well as the object's category and / or shape.

[0068] In this embodiment, the global top view may include at least one object, and each of the at least one object may be processed separately. Optionally, at least one object may be processed sequentially, or at least one object may be processed simultaneously.

[0069] Optionally, the object can be any object at the target accident scene, such as a vehicle, person, electric vehicle, road line, etc.

[0070] In one embodiment, at least one object included in the global top view can be identified by deep learning methods or other object recognition methods (e.g., connected component analysis). Optionally, the global top view can be semantically segmented by a semantic segmentation model to obtain a category label + object mask corresponding to at least one object. The category corresponding to the object can be directly determined based on the category label, and the shape and size of the object can be determined based on the shape and size of the mask. Optionally, by using connected component analysis, the global top view is first binarized to obtain a binarized top view. Neighborhood connected component detection is performed on the binarized top view to output a mask + bounding rectangle corresponding to each object. The size and shape of the object can be determined based on the mask and the bounding rectangle.

[0071] Step 1042: Determine the initial object symbol corresponding to the object based on the object's category and / or shape.

[0072] In this embodiment, the initial object symbol can be selected from a pre-created set of multiple preset object symbols. For example, the set includes preset object symbols with category names and corresponding illustrations, such as trucks, cars, electric vehicles, and human bodies. The preset object symbol corresponding to the category of the object can be determined as the initial object symbol. For example, if the category of the object is electric vehicle, the preset object symbol with the category name "electric vehicle" in the set can be directly obtained as the initial object symbol of the object. Alternatively, the shape of the object can be matched with the shapes of multiple preset object symbols to obtain the preset object symbol with the matching shape as the initial object symbol of the object. Optionally, when a unique initial object symbol cannot be determined based on either the category or the shape alone, it can be determined by combining the two. For example, when multiple preset object symbols are determined by searching based on the category, the shape is then matched with the multiple preset object symbols obtained by searching to determine the unique preset object symbol with the matching shape as the initial object symbol.

[0073] Optionally, step 1042 may include: determining the corresponding target symbol set based on the scene type corresponding to the target accident scene.

[0074] Each scene type corresponds to a symbol set, which includes preset object symbols for multiple different objects.

[0075] Different accident scenes correspond to different scenario types. For example, traffic accident scenes correspond to traffic scenarios, while public safety accident scenes correspond to public places (such as shopping malls, parks, etc.). The objects involved in the accident differ in different scenarios; therefore, corresponding symbol sets can be established for different scenarios. For example, the target symbol set corresponding to a traffic scenario can include, but is not limited to, trucks, cars, electric vehicles, etc.

[0076] Based on the object's category and / or shape, match it with multiple preset object symbols in the target symbol set, and determine the preset object symbol that matches the object's shape as the initial object symbol.

[0077] This embodiment can directly determine the corresponding initial object symbol for an object from the target symbol set by category. For example, after processing the global top view using a semantic segmentation network model, if a target object is determined to be a sedan, then a preset object symbol of category "sedan" is directly selected from the target symbol set as the initial object symbol for that target object. Alternatively, the shape of the target object can be matched with multiple preset object symbols in the target symbol set. Through matching (the matching process can be implemented through similarity calculation, deep neural networks, etc.), the object's shape is determined to be closest to that of a sedan, and the preset object symbol corresponding to the sedan is selected as the initial object symbol. Furthermore, both methods can be combined. For example, if the shape recognition shows that the object is similar to both a truck and a sedan, further matching by category can be performed to determine a more accurate symbol. This embodiment achieves fast and accurate determination of the initial object symbol by combining shape and / or category. Moreover, by matching from preset object symbols, the resulting initial object symbol has a standard shape, overcoming the non-standard and error problems caused by human drawing.

[0078] Step 1043: Adjust the size of the initial object symbol according to the size of the object to obtain the object symbol of at least one object in the global top view.

[0079] In this embodiment, the preset object symbols are usually set to a uniform size (smaller for easier display) for ease of management. However, this uniform size cannot fully match the object. Therefore, this embodiment adjusts the size of the initial object symbols based on the size of the target object to obtain a target symbol that matches the size of the object, thereby achieving a better standardized display of the object in the drawing.

[0080] In some alternative embodiments, after performing step 102, the following may also be included: Based on the point cloud data corresponding to at least one object, determine the location information and orientation of at least one object.

[0081] Optionally, after determining the initial object symbol, in order to accurately represent the object in the global top view, it is also necessary to determine the object's position information to achieve accurate insertion of the initial object symbol. Optionally, the position information and orientation of the object are determined based on the point cloud data corresponding to the object.

[0082] Optionally, step 1043 may include: Insert at least one initial object symbol into the global top view based on the position information of at least one object.

[0083] In this embodiment, the two-dimensional coordinate information projected onto the global top view can be determined based on the coordinate information corresponding to the point cloud data of the object. This two-dimensional coordinate information can be used as the position information. In addition, at the accident scene, the orientation of the object can play a significant role in the judgment of the accident. Therefore, it is also necessary to determine the orientation of the object. Optionally, the orientation of the object can be determined based on the shape of the object. For example, the front and rear of a vehicle have different shapes, and the vehicle orientation can be determined based on the shape difference.

[0084] At least one initial object symbol is stretched according to the orientation of at least one target object until the length of the initial object symbol matches the size of the target object, thus obtaining the object symbol.

[0085] In this embodiment, the preset object symbols are typically small in size for ease of display. Therefore, the initial object symbols determined solely based on shape and / or name matching are also usually small in size. To correctly display the target object, the size of the initial object symbols needs to be adjusted. This embodiment uses position control to accurately insert the initial object symbols and stretches them according to the object's orientation, achieving directional stretching of the initial object symbols. This makes the matching process between the stretched object symbols and the object's size more accurate, avoiding repeated stretching and contraction issues caused by insufficient or excessive stretching.

[0086] This embodiment realizes intelligent vectorized drawing of symbols based on point cloud base map. It adopts the "one-stroke drawing" method to effectively transform subjective drawing into accurate drawing based on objective data, ensuring the accuracy of the absolute position, shape and size of drawing elements, and realizing the standardization of drawing.

[0087] Figure 4 This is a flowchart illustrating the process of determining an object symbol in a method provided by another exemplary embodiment of this disclosure. For example... Figure 4 As shown above, in the above Figure 1 Based on the illustrated embodiment, step 104 may include the following steps: Step 1041: Identify the size of at least one object included in the global top view, as well as the object's category and / or shape.

[0088] Step 1042: Determine the initial object symbol corresponding to the object based on the object's category and / or shape.

[0089] Step 1043: Adjust the size of the initial object symbol according to the size of the object to obtain the object symbol of at least one object in the global top view.

[0090] The specific implementation and technical effects of steps 1041 to 1043 above can be found in [reference needed]. Figure 3 The provided embodiments are for illustrative purposes only and will not be elaborated upon further here.

[0091] Step 1044: Adjust the object symbol accordingly based on at least one received adjustment instruction to obtain the adjusted object symbol.

[0092] exist Figure 3 In the illustrated embodiment, the initial drawing of the object in the global top view is basically completed, achieving a standardized display of the object's basic shape and size in the global top view. However, the initially drawn object symbol may differ from the actual object, for example, there may be angular deviations or differences in aspect ratio. To solve the problem of mismatch between the object symbol and the object, this embodiment uses adjustment commands to make corresponding adjustments to the object symbol, so that the adjusted object symbol matches the object more accurately, enabling a more accurate description of the target accident scene and providing a basis for staff to handle the accident correctly.

[0093] Optionally, step 1044 may include: Receive at least one adjustment instruction, and determine at least one corresponding adjustment operation based on the at least one adjustment instruction.

[0094] Among them, at least one adjustment operation includes, but is not limited to, at least one of the following: rotation, translation, scaling, and changing aspect ratio; each adjustment instruction corresponds to one adjustment action.

[0095] Optionally, the adjustment instructions can be user-inputted or determined by a deep learning algorithm, for example, by receiving user-inputted adjustment instructions on a terminal (e.g., a tablet computer, etc.) that displays the global top view.

[0096] Perform at least one adjustment operation on the object symbol in a preset order to obtain the adjusted object symbol.

[0097] Optionally, the preset order can be determined based on the time of receiving the adjustment instruction, or based on the difficulty of the adjustment operation, for example, by arranging complex adjustments later.

[0098] In this embodiment, when at least one adjustment instruction is received, the instructions are received in a specific order. To avoid errors caused by performing multiple adjustment operations simultaneously, at least one adjustment operation can be performed on the object symbol in the order in which the instructions are received. Optionally, a corresponding timestamp can be assigned to each adjustment instruction when it is received. After obtaining the adjustment operation, at least one adjustment operation is sorted according to the timestamp, and the object symbol is adjusted sequentially according to the sorted at least one adjustment action to obtain the adjusted object symbol.

[0099] Figure 5 This is a schematic flowchart illustrating the process of determining reference elements in an exemplary embodiment of this disclosure. Figure 5 As shown above, in the above Figure 1 Based on the illustrated embodiment, step 106 may include the following steps: Step 1061: Determine the distribution of at least one object symbol in the global top view based on the position of at least one object symbol in the global top view.

[0100] Optionally, the global top-down view in this embodiment is obtained based on point cloud data. The three-dimensional coordinates of each object symbol are known in the point cloud data. Therefore, the two-dimensional coordinates of each object symbol can be known in the global top-down view corresponding to the point cloud data. Based on the two-dimensional coordinates, the position of each object symbol in the global top-down view can be determined. In this embodiment, distribution refers to the relative positional relationship between at least one object symbol and the distance of each object symbol from the edge in the global top-down view. For example, in a traffic accident scenario, lane lines are usually distributed at the edge of the global top-down view. Therefore, the distribution of each object symbol in the global top-down view can be determined based on its position.

[0101] Step 1062: Based on the distribution of at least one object symbol in the global top view, determine at least one object symbol located at a preset position in the distribution as at least one reference element.

[0102] In this embodiment, the reference elements are typically inherent facilities in the scene, possessing immovability and unaffected by external forces such as collisions or impacts during traffic accidents. Preferably, they are objects permanently fixed in the road scene, such as lane lines (solid center lines, edge lines, lane dividers, etc.), traffic light poles, traffic sign posts, stop lines, zebra crossings, curbs, guardrails, and bridge piers, excluding movable or easily damaged objects such as vehicles, temporary storage items, and trees prone to falling. Optionally, object symbols at specific locations (e.g., edge positions, middle positions, etc.) within the distribution are identified as reference elements; in other optional embodiments, which object symbols are used as reference elements can be determined based on user instructions, for example, the user clicking on lane lines and / or traffic lights as reference elements.

[0103] Figure 6This is a schematic flowchart illustrating the method for determining distance information provided in an exemplary embodiment of this disclosure. For example... Figure 6 As shown above, Figure 1 Based on the illustrated embodiment, a high-resolution image of at least one object at the target accident scene is obtained before or after performing step 102. Alternatively, a high-resolution image of at least one object at the target accident scene is obtained simultaneously with performing step 102. Each object corresponds to at least one high-resolution image.

[0104] Optionally, for the same accident scene, a laser SLAM device with RGB-D imaging capability (such as a laser SLAM scanner based on the Cartographer algorithm, which integrates a 16-line / 32-line LiDAR, a high-definition RGB camera and an IMU inertial measurement unit) is used. The surveyor can hold the device by hand or mount it on a mobile platform and move it around the accident scene at a constant speed along a preset path (for example, in a traffic accident scene, priority is given to covering the vehicle collision area, the area where brake marks are distributed, and the area around fixed road references). When the SLAM equipment is working, the lidar collects distance information of the scene environment in real time, the IMU synchronously collects the equipment's own attitude (pitch angle, roll angle, yaw angle) and motion parameters, and the high-definition RGB camera synchronously collects environmental color texture information. The equipment's built-in SLAM algorithm (such as Cartographer, LOAM) performs real-time fusion processing on the lidar point cloud, IMU data and RGB image, eliminates accumulated errors through closed-loop detection, and constructs a 1:1 high-precision color point cloud model of the accident scene in real time (point cloud density ≥1000 points / ㎡, position accuracy ≤±2cm), ensuring that the spatial position of all objects at the scene (vehicles, collision points, brake marks, lane lines, traffic light poles, etc.) is accurately mapped. During SLAM mobile scanning, high-definition images of the site are captured simultaneously, and the camera pose data at the time of image capture is recorded. All high-definition images are mounted under the same coordinate system. For example, when surveyors identify key details such as vehicle collision points, brake mark start and end points, and debris landing points through the device's built-in touch screen or external terminal, the device's high-definition close-up shooting function is triggered. The RGB camera (resolution ≥ 4K, adjustable focal length, aperture F2.8-F5.6) captures high-definition close-up photos of key details. At the same time, the device automatically records the camera pose (extrinsic parameters) and shooting parameters (focal length, aperture, exposure time) at the time of each close-up photo capture, and timestamps it with the real-time constructed color point cloud model to provide basic data support for subsequent coordinate transformation.

[0105] Step 108 may include the following steps: Step 1081: Based on the click command received in the high-definition image or the identification of the accident-related area in the high-definition image, determine the image coordinates of the accident key points in the object symbol in the global top view.

[0106] In this embodiment, the details of objects at the target accident scene can be displayed in the obtained high-definition image. The coordinates of key points related to the accident can be quickly determined through the details displayed in the high-definition image. After determining the coordinates of the key points in the high-definition image, the three-dimensional world coordinates corresponding to the key points can be determined through the correspondence between the high-definition image and point cloud data. Then, by combining the projection relationship between the point cloud data and the global top view, the image coordinates of the key points in the global top view can be determined. In some optional examples, step 1081 may include: Based on click commands received in the high-definition image or by identifying key points of the accident in the high-definition image, determine the coordinates of key points in the high-definition image.

[0107] Optionally, a click command can be received via the terminal screen to obtain the user's click position on the terminal screen. This position is initially a normalized coordinate in the screen coordinate system (coordinate range [0,1]×[0,1]), where the horizontal coordinate corresponds to the screen width percentage and the vertical coordinate corresponds to the screen height percentage. Based on the terminal screen resolution (e.g., 1920×1080) and the display scaling ratio of the close-up photo, the normalized coordinates are converted into pixel coordinates (u,v) of the original high-definition close-up photo using a linear mapping algorithm, where u is the pixel horizontal coordinate (unit: pixels) and v is the pixel vertical coordinate (unit: pixels). During the conversion process, coordinate deviations caused by screen scaling and stretching are eliminated to ensure that the pixel coordinates correspond one-to-one with the pixel positions of the original photo. Alternatively, a trained deep learning model (neural network model, etc.) can be used to perform key point recognition on the high-definition image. After identifying the key points of the accident, for example, through region of interest recognition technology, the identified coordinates are determined as key point coordinates.

[0108] Based on the correspondence between high-definition images and point cloud data, the coordinates of key points are transformed to determine the target three-dimensional coordinates of the key accident points in the point cloud data.

[0109] Optionally, a click command can be received via the terminal screen to obtain the user's click position on the terminal screen. This position is initially a normalized coordinate in the screen coordinate system (coordinate range [0,1]×[0,1]), where the horizontal coordinate corresponds to the screen width percentage and the vertical coordinate corresponds to the screen height percentage. Based on the terminal screen resolution (e.g., 1920×1080) and the display scaling ratio of the close-up photo, the normalized coordinates are converted into pixel coordinates (u,v) of the original high-definition close-up photo using a linear mapping algorithm, where u is the pixel horizontal coordinate (unit: pixels) and v is the pixel vertical coordinate (unit: pixels). During the conversion process, coordinate deviations caused by screen scaling and stretching are eliminated to ensure that the pixel coordinates correspond one-to-one with the pixel positions of the original photo. Alternatively, a trained deep learning model (neural network model, etc.) can be used to perform key point recognition on the high-definition image. After identifying the key points of the accident, the identified coordinates are determined as the key point coordinates.

[0110] In some optional examples, the process of determining the image coordinates of critical accident points in the target top view may include: The correspondence between high-resolution images and point cloud data is determined based on the pose information of the camera that acquired the high-resolution images.

[0111] This embodiment records the pose information of the acquisition camera in real time while acquiring high-definition images, providing a basis for realizing the back projection of 2D pixel coordinates into 3D camera spatial coordinates.

[0112] Based on the correspondence, the coordinates of key points are transformed to determine the target three-dimensional coordinates of the key accident points in the point cloud data.

[0113] Based on the projection relationship between point cloud data and the global top view, determine the image coordinates corresponding to the target's three-dimensional coordinates.

[0114] In this embodiment, the pre-calibrated camera intrinsic parameter matrix K of the SLAM device (the device that collects point cloud data) is first invoked (this intrinsic parameter matrix is ​​obtained in advance and includes the camera focal lengths f_x and f_y, principal point coordinates (c_x, c_y), and distortion coefficients). The intrinsic parameter matrix is ​​used to perform principal point offset elimination and focal length normalization processing to eliminate the influence of distortion (radial distortion and tangential distortion) of the camera optical system itself. Subsequently, based on the pinhole camera imaging model, the 2D pixel coordinates (u, v) are combined with the effective depth value Z obtained in the above steps and restored to 3D camera space coordinates (X_c, Y_c, Z_c) through a back projection algorithm, where X_c and Y_c are the horizontal and vertical coordinates of the camera space, and Z_c is the depth coordinate of the camera space. The back projection formula strictly follows the pinhole camera imaging principle to ensure accurate mapping between 2D pixels and 3D camera coordinates. 3D camera space coordinates are converted to global world coordinates: When the system calls the SLAM device to capture the close-up photo, it synchronously records the camera extrinsic parameter (pose) matrix T (containing rotation matrix R and translation vector t). The rotation matrix R describes the pose angle of the camera space relative to the global world space, and the translation vector t describes the position coordinates of the camera center in the global world space (this extrinsic parameter is calculated in real time by the SLAM algorithm, optimized by combining IMU data and point cloud matching results, with a position accuracy ≤ ±3cm). Through rotation and translation transformation algorithms, the obtained 3D camera space coordinates (X_c, Y_c, Z_c) are transformed to the global world space. During the transformation, the coordinate transformation is achieved through matrix multiplication (world coordinates = rotation matrix × camera coordinates + translation vector). Finally, the precise 3D world coordinates (X_w, Y_w, Z_w) of the clicked position are output. These coordinates are consistent with the coordinate system of the color point cloud model constructed by the SLAM device, which is the target 3D coordinates (3D world coordinates) in the point cloud data.

[0115] After determining the target's three-dimensional coordinates corresponding to the key points of the accident in the point cloud data, the image coordinates of the key points of the accident in the global top view are determined based on the projection relationship between the point cloud data and the global top view (e.g., setting the z-axis coordinate value to zero).

[0116] Step 1082: Determine the target reference feature from at least one reference feature based on the image coordinates of the key points of the accident.

[0117] In this embodiment, after determining the image coordinates of the key points of the accident, the nearest reference element can be automatically retrieved as the target reference element for the image coordinates, or a preset reference element can be determined as the target reference element according to the instruction (click / select, etc.). The key points of the accident can be described by reference by determining the target reference element.

[0118] Step 1083: Based on the positional relationship between the accident key points and the target reference elements, and the positional relationship between at least one accident key point, determine the annotation information of the accident key points.

[0119] Optionally, the annotation information may include, but is not limited to, at least one of the following: distance information between the accident key point and the target reference element, descriptive information of the accident key point (e.g., tire trail length, scratch mark length, etc.). The descriptive information of the accident key point can be determined by the positional relationship between two accident key points, which can be determined based on the coordinates of the two accident key points in the global top view. After connecting the two accident key points, the distance information between the two accident key points is calculated based on the coordinates, which can then be used as the descriptive information of the accident key point. For example, in an optional example, Figure 7 This is a schematic diagram showing the annotation information of at least one key accident point in the method provided in an optional example of Embodiment 1 of this disclosure. For example... Figure 7 As shown in the figure, key accident points such as the four wheels of the vehicle, scattered objects, a cola bottle, and bloodstains are displayed (the key accident points in the figure are used as measurement points). The lane line on one side of the vehicle is determined as the baseline, and the lamppost is determined as the reference point. Corresponding labeling information is assigned to these key accident points. The length of the left rear tire trail is determined based on the distance between the left rear tire and the tire track as two key accident points. In the example in the figure, the length of the left rear tire trail is 268 pixels.

[0120] After the above steps, the image coordinates of the key accident points and target reference elements in the global top view are determined; given two coordinates in the same coordinate system, the distance between the coordinates can be determined; the type of reference element may include, but is not limited to, reference points and / or reference lines. Optionally, step 1062 may include: Determine the type of target baseline element.

[0121] In response to the target reference element being the baseline, the annotation information is determined based on the vertical distance between the accident critical point and the baseline; and / or, in response to the target reference element being the reference point, the annotation information is determined based on the line segment length between the accident critical point and the reference point.

[0122] In this embodiment, due to the different types of reference elements, the methods for determining distance information also differ. When the reference element is a baseline, the distance from a point to a line segment is the perpendicular distance from that point to the line segment. Since the key point of the accident is a single point, the distance from the key point of the accident to the baseline can be determined by drawing a perpendicular line from the key point of the accident to the baseline, obtaining the foot of the perpendicular, and then determining the distance from the key point of the accident to the foot of the perpendicular as the distance information from the key point of the accident to the baseline. When the reference element is a reference point, the distance between two points can be directly determined based on the line connecting the two points. This embodiment associates the key point of the accident with at least one reference element through distance information and realizes the attribute description relative to at least one reference element. When there are multiple key points of the accident in the global top view (which may correspond to multiple objects or the same object), each key point of the accident can be described separately based on the distance information of the same reference element corresponding to each key point of the accident. This can achieve a more intuitive and unified description of the key points of the accident, providing a data foundation for subsequent analysis and processing of the target accident scene without requiring too many calculation steps.

[0123] Figure 8 This is a schematic flowchart illustrating the process of annotating key points of an accident in an exemplary embodiment of this disclosure. For example... Figure 8 As shown, based on any of the above embodiments, the annotation information includes attribute information between key accident points and distance information between key accident points and reference features. Step 110 may include: Step 1101: Based on the positional relationship between at least one key point of the accident, establish a key point association line segment, and display the attribute information between the key points of the accident on the key point association line segment.

[0124] As mentioned above Figure 7 As shown in the embodiment, describing key points of an accident requires not only the distance information between the key points and baseline elements, but also sometimes the description in conjunction with other key points. For example, the length of a tire track needs to be determined by combining the tire track key point and the tire track key point (the distance between them), etc. This embodiment achieves the attribute description of items related to the accident through attribute information annotation, enabling a more intuitive understanding of the attributes of accident-related items in the drawn vector diagram, and achieving a more accurate quantitative representation of the accident scene.

[0125] Step 1102: Establish the marker line segments from the key accident points to the corresponding benchmark elements, and display the annotation information corresponding to the key accident points on the marker line segments to obtain the on-site vector map.

[0126] Based on at least one distance information, marker line segments are established between the critical accident points and the baseline elements. Each marker line segment corresponds to one distance information.

[0127] When the reference element is a reference point, the two points are directly connected to form a marker line segment. When the reference element is a reference line, a perpendicular line segment is drawn from the accident critical point to the reference line, and the shortest distance line segment from the accident critical point to the reference line is established with the foot of the perpendicular as the midpoint. The marker line segment explicitly links the accident critical point with the reference element, making the relative positions of vehicles, traces, and objects on the road visible and geometric, avoiding the abstractness of pure text or pure coordinate descriptions.

[0128] Display the corresponding distance information on the marked line segments to complete the distance marking of key accident points.

[0129] In this embodiment, the determined key point association line segments and marker line segments serve as the geometric carriers of attribute information and distance information, which can intuitively represent the spatial relationships between key accident points and between key accident points and benchmark elements. Optionally, attribute information can be marked on one side of the key point association line segment or separately marked through indicator signs, and the obtained distance information can be marked in real time next to or in the middle of the marker line segment, forming a measurement label with numerical values, realizing an integrated presentation of geometric relationships and quantitative values. The calculated precise distance is marked on the corresponding marker line segment, and surveyors and appraisers can directly read the values ​​without secondary calculation, greatly improving the efficiency of on-site analysis and recording. In addition, based on a unified coordinate system, the marker line segments and distance labels are automatically generated by coordinates, unaffected by human perspective or subjective judgment, ensuring that the measurement results are reproducible, verifiable, and cross-verifiable. Figure 9 This is a schematic diagram of a vector graphic generated in a method provided in another exemplary embodiment of this disclosure. For example... Figure 9As shown, the reference elements in this embodiment include baselines and reference points. The illustrated embodiment includes multiple accident key points, which are used to describe the target accident scene. In the figure, the car is symbolized and displayed as a standard car symbol (size, position, and orientation match the target accident scene), and the bicycle is symbolized and displayed as a standard bicycle symbol (size, position, and orientation match the target accident scene). By using the four tires of the car and their corresponding tire tracks as accident key points, the drag mark length of each tire can be obtained. Based on the drag mark length, accident-related personnel (or analysis algorithms, etc.) can analyze whether the car braked before the accident, and thus analyze the nature of the accident. Furthermore, by marking the distance information between each accident key point and the baseline, the positional relationship between accident-related items can be determined more accurately, ensuring data consistency under the same reference object. For example, the distance information between the wheel and the baseline is marked in the middle of the marked line segment, completing the marking of an accident key point in the global top view. Dimension lines and values ​​conforming to national standards are automatically generated on the layer, and finally, a 1:1 scene map is formed based on the point cloud. Optionally, at the scene of a traffic accident, after generating a vectorized scene map, the parties involved in the accident need to complete an electronic signature confirmation on a display terminal (e.g., a tablet computer) to generate an unalterable electronic case file and upload it to the traffic management platform, thereby achieving on-site investigation and mapping within minutes.

[0130] The method for drawing an accident scene vector map provided in this disclosure can be executed by any suitable device with data processing capabilities, including but not limited to terminal devices and servers. Alternatively, the method for drawing an accident scene vector map provided in this disclosure can be executed by a processor, such as by a processor calling corresponding instructions stored in memory to execute any of the accident scene vector map drawing methods mentioned in this disclosure. Further details will not be elaborated below.

[0131] Exemplary device Figure 10 This is a schematic diagram of the structure of an apparatus for drawing a vector map of an accident scene provided in an exemplary embodiment of this disclosure. Figure 10 As shown, the apparatus provided in this embodiment includes: The object recognition module 11 is used to perform object recognition on the global top view of the target accident scene and determine that the target accident scene includes at least one object; The object symbolization module 12 is used to symbolize at least one object in the global top view to obtain at least one object symbol; each object symbol corresponds to one object. The benchmark determination module 13 is used to determine at least one benchmark element based on the distribution of the at least one object symbol in the global top view; The annotation information module 14 is used to determine the annotation information of at least one accident key point in the global top view based on the at least one reference element; each accident key point corresponds to one object; Vectorization module 15 is used to obtain a symbolic on-site vector map with annotation information based on the annotation information of the at least one key accident point and the at least one object symbol.

[0132] The apparatus provided in the above embodiments of this disclosure performs object recognition on a global top-down view of a target accident scene to determine at least one object included in the target accident scene; it enables rapid and automatic acquisition of objects at the accident scene, providing a basis for subsequent object drawing; it performs symbolization processing on at least one object in the global top-down view to obtain at least one object symbol; object symbolization achieves the standardization of drawing; based on the distribution of the at least one object in the global top-down view, it determines at least one reference element; by determining at least one reference element in the global top-down view, it provides a reference object for object annotation, and by referring to the object, it can achieve unified standard annotation of at least one object; based on the at least one reference element, it determines the annotation information of at least one key accident point in the global top-down view; based on the annotation information of the at least one key accident point and the at least one object symbol, it obtains a symbolized scene vector map with annotation information. This disclosure embodiment determines the key points of the accident for key points related to the target accident (e.g., collision points), and marks the key points of the accident in combination with reference elements, thereby realizing automatic annotation of the target accident scene. This avoids drawing errors caused by manual drawing and annotation, and solves the problem that it is impossible to annotate the accident scene in detail by relying solely on point cloud data. Furthermore, by combining symbolic object symbols, it realizes the automatic drawing of the standard vector map of the target accident scene while automatically annotating the details of the accident scene.

[0133] In some optional embodiments, the reference determination module 13 is specifically used to determine the distribution of at least one object symbol in the global top view based on the position of at least one object symbol in the global top view; and to determine at least one object symbol located at a preset position in the distribution as at least one reference element based on the distribution of at least one object symbol in the global top view.

[0134] In some optional embodiments, the apparatus provided in this disclosure may further include: The image acquisition module is used to acquire high-resolution images of at least one object; each object corresponds to at least one high-resolution image; The annotation information module 14 includes: The coordinate determination unit is used to determine the image coordinates of key accident points in the object in the global top view based on click instructions received in the high-definition image or by identifying accident-related areas in the high-definition image. The feature filtering unit is used to determine the target reference feature from at least one reference feature based on the image coordinates of the key points of the accident. The standard determination unit is used to determine the annotation information of the accident key points based on the positional relationship between the accident key points and the target reference elements, as well as the positional relationship between at least one accident key point.

[0135] Optionally, the standard determination unit is specifically used to determine the type of the target reference element; the type of reference element includes reference points and / or reference lines; in response to the target reference element being a reference line, annotation information is determined based on the vertical distance between the accident critical point and the reference line; and / or, in response to the target reference element being a reference point, annotation information is determined based on the line segment length between the accident critical point and the reference point.

[0136] Optionally, the coordinate determination unit is specifically used to determine the coordinates of key points in the high-definition image based on click instructions received in the high-definition image or identification of key accident points in the high-definition image; to transform the coordinates of key points based on the correspondence between the high-definition image and point cloud data to determine the target three-dimensional coordinates corresponding to the key accident points in the point cloud data; and to determine the image coordinates corresponding to the target three-dimensional coordinates based on the projection relationship between the point cloud data and the global top view.

[0137] In some optional embodiments, the vectorization module 15 is specifically used to establish key point association line segments based on the positional relationship between at least one key accident point, and display the attribute information between the key accident points on the key point association line segments; establish identification line segments from the key accident points to the corresponding reference elements, and display the annotation information corresponding to the key accident points on the identification line segments, thereby obtaining a field vector map.

[0138] In some optional embodiments, the object symbolization module 12 is specifically used to identify the size of at least one object included in the global top view, as well as the object's category and / or shape; determine the initial object symbol corresponding to the object based on the object's category and / or shape; and adjust the size of the initial object symbol according to the object's size to obtain the object symbol of at least one object in the global top view.

[0139] Optionally, the object symbolization module 12 is further configured to determine the position information and orientation of at least one object based on the point cloud data corresponding to at least one object; When the object symbolization module 12 adjusts the size of the initial object symbol according to the size of the object to obtain the object symbol of at least one object in the global top view, it inserts at least one initial object symbol into the global top view according to the position information of at least one object; and performs stretching processing on at least one initial object symbol according to the orientation of at least one target object until the length of the initial object symbol matches the size of the target object to obtain the object symbol.

[0140] In some optional embodiments, the object recognition module 11 is specifically used to determine a global top view based on point cloud data collected from the target accident scene; identify at least one region of interest in the global top view; and determine at least one object included in the target accident scene.

[0141] The exemplary embodiment of the apparatus for drawing accident scene vector maps provided in this disclosure corresponds to the above-described exemplary method for drawing accident scene vector maps in terms of implementation. The corresponding content of the two can be referenced, combined, and cited from each other, and will not be repeated here. The beneficial technical effects corresponding to the exemplary embodiment of the apparatus for drawing accident scene vector maps provided in this disclosure can be found in the corresponding beneficial technical effects of the above-described exemplary method for drawing accident scene vector maps, and will not be repeated here.

[0142] Exemplary electronic devices Below, for reference Figure 11 This describes an electronic device according to embodiments of the present disclosure. The electronic device may be either or both of a first device and a second device, or a standalone device independent of them, which may communicate with the first device and the second device to receive acquired input signals from them.

[0143] Figure 11 A block diagram of an electronic device according to an embodiment of the present disclosure is shown.

[0144] like Figure 11 As shown, the electronic device includes one or more processors and memory.

[0145] A processor can be a central processing unit (CPU) or other form of processing unit with data processing and / or instruction execution capabilities, and can control other components in an electronic device to perform desired functions.

[0146] The memory can store one or more computer program products, and the memory can include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program products can be stored on the computer-readable storage medium, and the processor can run the computer program products to implement the methods for drawing accident scene vector diagrams according to the various embodiments of this disclosure described above, and / or other desired functions.

[0147] In one example, the electronic device may also include input devices and output devices, which are interconnected via a bus system and / or other forms of connection mechanism (not shown).

[0148] In addition, the input device may also include, for example, a keyboard, a mouse, etc.

[0149] This output device can output various information to the outside, including determined distance information, direction information, etc. The output device may include, for example, a display, a speaker, a printer, and a communication network and its connected remote output devices, etc.

[0150] Of course, for the sake of simplicity, Figure 11 Only some of the components of the electronic device relevant to this disclosure are shown, omitting components such as buses, input / output interfaces, etc. In addition, the electronic device may include any other suitable components depending on the specific application.

[0151] In addition to the methods and devices described above, embodiments of this disclosure may also be computer program products, including computer program instructions that, when executed by a processor, cause the processor to perform the steps in the methods for drawing accident scene vector diagrams according to various embodiments of this disclosure as described in the foregoing portion of this specification.

[0152] The computer program product can be written in any combination of one or more programming languages ​​to perform the operations of the embodiments of this disclosure. The programming languages ​​include object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on a user's computing device, partially on a user's computing device, as a standalone software package, partially on a user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0153] Furthermore, embodiments of this disclosure may also be computer-readable storage media storing computer program instructions thereon, which, when executed by a processor, cause the processor to perform the steps in the methods for drawing accident scene vector diagrams according to various embodiments of this disclosure as described in the foregoing portion of this specification.

[0154] The computer-readable storage medium may be any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof.

[0155] The basic principles of this disclosure have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this disclosure are merely examples and not limitations, and should not be considered as essential features of each embodiment of this disclosure. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the scope of this disclosure to the necessity of employing the aforementioned specific details for implementation.

[0156] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For system embodiments, since they largely correspond to method embodiments, the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0157] The block diagrams of devices, apparatuses, devices, and systems disclosed herein are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0158] The methods and apparatus of this disclosure may be implemented in many ways. For example, they may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above-described order of steps for the methods is for illustrative purposes only, and the steps of the methods of this disclosure are not limited to the order specifically described above unless otherwise specifically stated. Furthermore, in some embodiments, this disclosure may also be implemented as a program recorded on a recording medium, the program including machine-readable instructions for implementing the methods according to this disclosure. Thus, this disclosure also covers recording media storing programs for performing the methods according to this disclosure.

[0159] It should also be noted that in the apparatus, devices, and methods of this disclosure, the components or steps can be disassembled and / or recombined. These disassemblies and / or recombinations should be considered as equivalent solutions to this disclosure.

[0160] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of this disclosure. Therefore, this disclosure is not intended to be limited to the aspects shown herein, but rather to be carried out within the widest scope consistent with the principles and novel features disclosed herein.

[0161] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this disclosure to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations therein.

Claims

1. A method for drawing a vector diagram of an accident scene, characterized in that, include: Object identification is performed on a global top-down view of the target accident scene to determine at least one object included in the target accident scene; At least one object in the global top view is symbolized to obtain at least one object symbol; each object symbol corresponds to one object. Based on the distribution of the at least one object symbol in the global top view, at least one reference element is determined; Based on the at least one reference element, determine the annotation information of at least one accident key point in the global top view; Each of the aforementioned key points of the accident corresponds to one of the aforementioned objects; Based on the annotation information of the at least one key accident point and the at least one object symbol, a symbolic on-site vector map with annotation information is obtained.

2. The method according to claim 1, characterized in that, The determination of at least one reference feature based on the distribution of the at least one object symbol in the global top view includes: The distribution of the at least one object symbol in the global top view is determined based on the position of the at least one object symbol in the global top view; Based on the distribution of the at least one object symbol in the global top view, the at least one object symbol located at a preset position in the distribution is determined as the at least one reference element.

3. The method according to claim 1 or 2, characterized in that, Also includes: Acquire a high-resolution image of the at least one object; Each of the objects corresponds to at least one of the high-definition images; The step of determining the annotation information of at least one accident key point in the global top view based on the at least one reference feature includes: Based on the click command received in the high-definition image or by identifying the accident-related area in the high-definition image, determine the image coordinates of the accident key points in the object in the global top view; The target reference element is determined from the at least one reference element based on the image coordinates of the key points of the accident; Based on the positional relationship between the accident key point and the target reference element, and the positional relationship between the at least one accident key point, the annotation information of the accident key point is determined.

4. The method according to claim 3, characterized in that, The step of determining the annotation information of the key accident points based on the positional relationship between the key accident points and the target reference elements includes: Determine the type of the target reference feature; the type of the reference feature includes reference points and / or reference lines; In response to the target reference element being the reference line, the annotation information is determined based on the vertical distance between the accident critical point and the reference line; and / or, In response to the target reference element being the reference point, the annotation information is determined based on the line segment length between the accident critical point and the reference point.

5. The method according to claim 3 or 4, characterized in that, The step of determining the image coordinates of key accident points in the object in the global top view based on click commands received in the high-definition image or by identifying accident-related areas in the high-definition image includes: Based on the click command received in the high-definition image or by identifying key points of the accident in the high-definition image, determine the coordinates of key points in the high-definition image; Based on the correspondence between the high-definition image and the point cloud data, the coordinates of the key points are transformed to determine the target three-dimensional coordinates of the key accident points in the point cloud data. Based on the projection relationship between the point cloud data and the global top view, the image coordinates corresponding to the three-dimensional coordinates of the target are determined.

6. The method according to any one of claims 1-5, characterized in that, The process of obtaining a symbolized on-site vector map with annotation information based on the annotation information of at least one key accident point and the at least one object symbol includes: Based on the positional relationship between the at least one key accident point, establish a key point association line segment, and display the attribute information between the key accident points on the key point association line segment; Establish marker line segments from the key accident points to the corresponding reference elements, and display the annotation information corresponding to the key accident points on the marker line segments to obtain the on-site vector map.

7. The method according to any one of claims 1-6, characterized in that, The symbolization process for at least one object in the global top view to obtain at least one object symbol includes: Identify the size of at least one object included in the global top view, as well as the category and / or shape of the object; Determine the initial object symbol corresponding to the object based on the object's category and / or shape; The size of the initial object symbol is adjusted according to the size of the object to obtain the object symbol of the at least one object in the global top view.

8. The method according to claim 7, characterized in that, After performing object identification on the global top-down view of the target accident scene to determine that the target accident scene includes at least one object, the method further includes: Based on the point cloud data corresponding to the at least one object, determine the position information and orientation of the at least one object; The step of adjusting the size of the initial object symbol according to the size of the object to obtain the object symbol of the at least one object in the global top view includes: According to the position information of the at least one object, insert the at least one initial object symbol into the global top view; The at least one initial object symbol is stretched according to the orientation of the at least one target object until the length of the initial object symbol matches the size of the target object, thus obtaining the object symbol.

9. The method according to any one of claims 1-8, characterized in that, The step of performing object identification on a global top-down view of the target accident scene to determine that the target accident scene includes at least one object, including: The global top view is determined based on the point cloud data collected from the target accident site; Identify at least one region of interest in the global top view and determine at least one object included in the target accident scene.

10. A device for drawing vector diagrams of accident scenes, characterized in that, include: The object recognition module is used to perform object recognition on a global top view of the target accident scene and determine that the target accident scene includes at least one object. An object symbolization module is used to symbolize at least one object in the global top view to obtain at least one object symbol; each object symbol corresponds to one object. A benchmark determination module is used to determine at least one benchmark element based on the distribution of the at least one object symbol in the global top view; The annotation information module is used to determine the annotation information of at least one accident key point in the global top view based on the at least one reference element; each accident key point corresponds to one object; The vectorization module is used to obtain a symbolic on-site vector map with annotation information based on the annotation information of the at least one key accident point and the at least one object symbol.

Images