A radar plot visualizing labeling method based on multi-feature view

CN122676006APending Publication Date: 2026-09-01NANJING RES INST OF ELECTRONICS TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202610850397.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-06-12
Publication Date
2026-09-01

AI Technical Summary

Technical Problem

传统的标注方法通常是在距离-方位角视图上对点迹进行可视化和人工标记,标注人员无法从多个维度直观地观察和区分不同特性的点迹

Benefits of technology

1.本发明提供了一种自由选择维度的视图显示方式,允许用户自由选择X轴和Y轴对应的点迹特征维度,用户可以在不同视图中全面分析雷达点迹数据,从而更深入地理解数据的多维特性。

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122676006A_ABST
    Figure CN122676006A_ABST
Patent Text Reader

Abstract

This invention discloses a radar point visualization annotation method based on multi-feature views, belonging to the field of information visualization. This method loads multi-dimensional radar point data to generate global and local views with freely definable X-axis and Y-axis feature dimensions; it provides two interactive modes: cursor selection and brush selection, along with corresponding deselection operations; it synchronizes the display status of points across all views and unifies label colors; and it supports annotation operations such as label assignment, label removal, point removal, and label creation. This invention solves the problems of traditional radar point annotation methods, such as single-dimensionality, low interactive efficiency, and difficulty in large-scale data annotation. It helps annotators intuitively observe and distinguish points from multiple dimensions, significantly improving the efficiency and accuracy of radar point data annotation and accelerating the research and development of artificial intelligence technologies related to radar target recognition.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of information visualization technology, and more specifically, to a radar point visualization and annotation method based on multi-feature views. Background Technology

[0002] With the widespread application of artificial intelligence technology in radar target recognition, classification, and tracking, the demand for high-quality labeled data is increasing. Radar detection data typically contains multiple feature dimensions, such as range, azimuth, elevation, and signal-to-noise ratio. Traditional labeling methods usually involve visualizing and manually marking points on a range-azimuth view, making it impossible for labelers to intuitively observe and distinguish points with different characteristics from multiple dimensions. Existing labeling tools are limited in function, typically only allowing selection and manipulation of points by clicking a single point or selecting a rectangular area, resulting in low efficiency and a high risk of errors. Furthermore, the large number of points can lead to overly dense visualizations, making it difficult to effectively display and select points accumulated over a long period, thus making the manual labeling of large-scale data tedious and difficult.

[0003] Therefore, there is an urgent need in this field for a visualization annotation scheme that can fully display the multi-dimensional features of radar points and provide flexible interactive means to improve annotation efficiency and accuracy. Summary of the Invention

[0004] This invention aims to overcome the shortcomings of the existing technology and provide an interactive visualization annotation method and system for radar points based on multiple feature views. Its core idea is to provide users with global, multi-perspective data insights by creating multiple scatter plot views with freely definable feature dimensions, supplemented by rich and intuitive interactive operations, greatly improving the efficiency and experience of data annotation.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: a radar point visualization and annotation method based on multi-feature views, characterized by comprising the following steps: S1: Data Loading: The system loads radar spot data containing multi-dimensional features and stores it in memory with a fixed data structure; S2: Multi-view construction: The user interface generates at least one global view and at least one local view; for each view window, the user can freely select the feature dimensions of the points corresponding to the X and Y axes, and the system draws the corresponding points in the coordinate system of the view in the form of scattered points. The global view displays all loaded points, and the local view only displays the points that the user is interested in. S3: Interaction Mode Selection: Users select the dot pattern interaction mode by clicking a button. The available modes include cursor selection and brush selection. S4: Dot selection operation: Performs dot selection based on the selected interaction mode, and the selected dots are highlighted synchronously in all views; S5: View zoom operation: Users can zoom the display area of ​​any view using the mouse wheel to make the target point clear. S6: Dot deselection operation: Performs dot deselection according to the current interaction mode, and the deselected dots are synchronously restored to their original display form in all views; S7: Label Management and Annotation Operations: Users can perform label assignment, label removal, or mark removal operations on the selected set of dots. All operation results are updated synchronously in each view. S8: Label Creation Operation: Users can create new labels using the new label function and assign them unique names and display colors.

[0006] In a preferred embodiment of the present invention, the dot selection operation for different interaction modes in step S4 specifically includes: Cursor selection mode: Select a single point by clicking on a single point with the mouse, or select all points within the rectangle by dragging the mouse. Brush selection mode: Move the mouse to a preset-sized circular brush area, press the left mouse button to start brush selection, and select the dots in the area where the brush passes. Release the left mouse button to stop brush selection.

[0007] In a preferred embodiment of the present invention, the deselection operation for different interaction modes in step S6 specifically involves: Cursor selection mode: The user can right-click in a blank area of ​​the view to deselect all selected points at once; Brush selection mode: Press the right mouse button to start cancel brush selection. Selected points in the area brushes across will be deselected. Release the right mouse button to stop the cancellation operation.

[0008] In a preferred embodiment of the present invention, the label management and annotation operation in step S7 specifically includes: Label Assignment: Assigns the selected label from the label list to the selected dot, and the dot's display color in all views is synchronously changed to the color corresponding to that label; Label Removal: Removes the label from the selected dot, and the dot is synchronously restored to its default color in all views; Dot Removal: Removes the selected dots from all views and makes them no longer displayed.

[0009] As a preferred embodiment of the present invention, the system establishes and maintains a tag list, which displays the names and corresponding display colors of all created tags, and the user selects a tag from the tag list as the tag to be assigned.

[0010] As a preferred embodiment of the present invention, each dot is drawn using a uniform color in all views, the color being determined by the label currently bound to the dot, while unlabeled dots use the system default color.

[0011] As a preferred embodiment of the present invention, when a user performs a dot selection, deselection, label assignment, label removal, or dot removal operation in any view, the system automatically updates the operation results to all other views in real time, thereby achieving real-time synchronization of multi-view states.

[0012] As a preferred embodiment of the present invention, the feature dimensions of the radar spot data include, but are not limited to, spot formation time, detection azimuth angle, detection range, detection elevation angle, spot energy amplitude, and signal-to-noise ratio.

[0013] In a preferred embodiment of the present invention, in step S4, after the dot is selected, it is highlighted in all views by enlarging its size, changing its color, or adding a border.

[0014] In a preferred embodiment of the present invention, in step S2, the user can dynamically add or delete view windows as needed, and adjust the X-axis and Y-axis feature dimensions of any view at any time.

[0015] This invention provides a method for visually annotating radar detection points, which has the following beneficial effects: 1. This invention provides a view display method with freely selectable dimensions, allowing users to freely select the feature dimensions of the point traces corresponding to the X and Y axes. Users can comprehensively analyze radar point trace data in different views, thereby gaining a deeper understanding of the multidimensional characteristics of the data.

[0016] 2. This invention provides a multi-view synchronous interactive method, in which each point is drawn using a uniform color determined by its current label in different views, and the display status is synchronized across all views. This function enables users to operate seamlessly between different views, improving the consistency and efficiency of data analysis.

[0017] 3. This invention can effectively improve the annotation efficiency of point data. Through multi-dimensional data visualization, flexible interaction mode, efficient label management and real-time feedback function, it greatly simplifies the user's analysis and processing process of radar detection point data and accelerates the research and development efficiency of related artificial intelligence technologies. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the method described in this invention. Figure 2 This is a schematic diagram of the system interface shown in an embodiment of the present invention. Detailed Implementation

[0019] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0020] This invention proposes a method for visually annotating radar detection points, such as... Figure 1 As shown, it includes the following steps: After the user starts the system, it initializes and creates four views: three global views and one local view. Each view contains two dropdown lists for selecting the feature dimensions corresponding to the X and Y axes, respectively. The overall system interface is as follows: Figure 2 As shown.

[0021] Users can set the feature dimensions corresponding to the X and Y axes of each view through drop-down lists. After the dimensions are set, the global view... Figure 1 The X-axis corresponds to the azimuth feature, the Y-axis corresponds to the distance feature, and the global view... Figure 2 The X-axis corresponds to the time feature, and the Y-axis corresponds to the pitch angle feature. The X-axis of the global view 3 corresponds to the time feature, and the Y-axis corresponds to the signal-to-noise ratio feature. The X-axis of the local view corresponds to the time feature, and the Y-axis corresponds to the distance feature.

[0022] Load a radar spot data file in .csv format. The file stores the features of spot detected by the radar over a period of time in the form of a two-dimensional list. Each row represents a spot, and each column represents a one-dimensional feature. The spot data contains a total of 6 columns, and the corresponding features are spot formation time, detection azimuth angle, detection range, detection elevation angle, spot energy amplitude, and signal-to-noise ratio.

[0023] After parsing, the dot data is displayed in various global views. Each dot in each view is represented by its two-dimensional position coordinates, which are based on the feature values ​​corresponding to its X and Y axes, and is displayed as a dot. Unlabeled dots are displayed in blue, while labeled dots are displayed in the color corresponding to their labels.

[0024] In the azimuth-distance view, the user sees a range that appears to contain a continuous trajectory, possibly a sequence of points representing a moving target. The user then clicks the cursor button to enter the cursor selection mode and selects the points in this area by dragging a box. The selected points are displayed as red dots that are twice the size in all views, and the selected points are also displayed separately in the local view.

[0025] In the partial view, the user sees a sequence of points within the selected points that shows a continuous change in distance over time. Assuming these points belong to the same target, the user right-clicks in a blank area to deselect them. Then, clicking the brush button enters brush selection mode, and the user uses the left mouse button to reselect these points with spatiotemporal continuity, along with some of their neighboring points, in the global view.

[0026] Users can observe and analyze selected points in various views, zoom in and out of the view area using the mouse wheel, and deselect unwanted points using the right mouse button in brush selection mode.

[0027] Users need to assign a unique batch number to these selected dots. The batch number label is based on the target, and all dots within each target should belong to the same batch number. When the user clicks the "New Label" button, the system automatically generates a new batch number and corresponding color, and adds it to the label list.

[0028] Users select the desired label from the label list and then click the "Assign Label" button to assign the corresponding label to the selected dots. After being labeled, the dots revert to their unselected state, and their color changes to the color of the corresponding label.

[0029] After assigning a label, the user finds that a dot is incorrectly labeled. So, the user enters the cursor selection mode, clicks to select the dot, and then clicks the label removal button to remove the label assigned to the dot.

[0030] The user observed what appeared to be a continuous trajectory within another area on the azimuth-distance view. Selecting this trajectory in brush selection mode revealed that while the points did exhibit some spatiotemporal continuity in the local view, they also showed significant jitter and were short, making it impossible to determine if they were genuine points. To prevent these points from negatively impacting model training, the user clicked the "Remove Points" button to remove them.

[0031] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A radar point visualization and annotation method based on multi-feature views, characterized in that, Includes the following steps: S1: Data Loading: The system loads radar spot data containing multi-dimensional features and stores it in memory with a fixed data structure; S2: Multi-view construction: The user interface generates at least one global view and at least one local view; for each view window, the user can freely select the feature dimensions of the points corresponding to the X and Y axes, and the system draws the corresponding points in the coordinate system of the view in the form of scattered points. The global view displays all loaded points, and the local view only displays the points that the user is interested in. S3: Interaction Mode Selection: Users select the dot pattern interaction mode by clicking a button. The available modes include cursor selection and brush selection. S4: Dot selection operation: Performs dot selection based on the selected interaction mode, and the selected dots are highlighted synchronously in all views; S5: View zoom operation: Users can zoom the display area of ​​any view using the mouse wheel to make the target point clear. S6: Dot deselection operation: Performs dot deselection according to the current interaction mode, and the deselected dots are synchronously restored to their original display form in all views; S7: Label Management and Annotation Operations: Users can perform label assignment, label removal, or mark removal operations on the selected set of dots. All operation results are updated synchronously in each view. S8: Label Creation Operation: Users can create new labels using the new label function and assign them unique names and display colors.

2. The radar point visualization and annotation method based on multiple feature views according to claim 1, characterized in that, In step S4, the point selection operation for different interaction modes is specifically as follows: Cursor selection mode: Select a single point by clicking on a single point with the mouse, or select all points within the rectangle by dragging the mouse. Brush selection mode: Move the mouse to a preset-sized circular brush area, press the left mouse button to start brush selection, and select the dots in the area where the brush passes. Release the left mouse button to stop brush selection.

3. The radar point visualization and annotation method based on multiple feature views according to claim 1, characterized in that, In step S6, the deselection operation for dots in different interaction modes is specifically as follows: Cursor selection mode: The user can right-click in a blank area of ​​the view to deselect all selected points at once; Brush selection mode: Press the right mouse button to start cancel brush selection. Selected points in the area brushes across will be deselected. Release the right mouse button to stop the cancellation operation.

4. The radar point visualization and annotation method based on multiple feature views according to claim 1, characterized in that, In step S7, the label management and annotation operation specifically includes: Label Assignment: Assigns the selected label from the label list to the selected dot, and the dot's display color in all views is synchronously changed to the color corresponding to that label; Label Removal: Removes the label from the selected dot, and the dot is synchronously restored to its default color in all views; Dot Removal: Removes the selected dots from all views and makes them no longer displayed.

5. The radar point visualization and annotation method based on multiple feature views according to claim 1, characterized in that, The system establishes and maintains a tag list, which displays the names and corresponding display colors of all created tags. The user selects a tag from the tag list as the tag to be assigned.

6. The radar point visualization and annotation method based on multiple feature views according to claim 1, characterized in that, Each dot is drawn using a uniform color across all views, determined by the label currently attached to that dot. Dots without labels use the system default color.

7. The radar point visualization and annotation method based on multiple feature views according to claim 1, characterized in that, When a user performs a dot selection, deselection, label assignment, label removal, or dot removal operation in any view, the system automatically updates the operation results to all other views, achieving real-time synchronization of multi-view states.

8. The radar point visualization and annotation method based on multiple feature views according to claim 1, characterized in that, The feature dimensions of the radar spot data include, but are not limited to, spot formation time, detection azimuth, detection range, detection elevation angle, spot energy amplitude, and signal-to-noise ratio.

9. The radar point visualization and annotation method based on multiple feature views according to claim 1, characterized in that, In step S4, once the dot is selected, it is highlighted in all views by enlarging its size, changing its color, or adding a border.

10. The radar point visualization and annotation method based on multiple feature views according to claim 1, characterized in that, In step S2, users can dynamically add or delete view windows as needed, and adjust the X-axis and Y-axis feature dimensions of any view at any time.