Real-time detection method and device for radar detection target and spatial position relationship

Abstract

The application discloses a real-time detection method and device for radar target detection and spatial position relationship, wherein the detection method comprises the following steps: dividing the detection plane of the radar into a spatial set in the form of a grid, each grid containing the position coordinates of the current grid and the position coordinates of the adjacent grid; updating the signals obtained by the radar in real time into the corresponding grid according to the coordinate values, wherein the information of the grid comprises the number of signal track points in the grid, the time stamp of the signal track points in the grid and the grid activity value; superimposing the spatial set and a set of time frames to form three-dimensional data; extracting a feature signal set and converting the feature signal set into a continuous coordinate curve.

Description

Technical Field

[0001] This invention relates to the field of surveillance, and more specifically, to a real-time detection method and device for radar-detected target-spatial relationship. Background Technology

[0002] As civilian millimeter-wave radar technology matures, it is being widely used in fields such as autonomous driving and perimeter security. In the field of perimeter security, millimeter-wave radar has unique advantages in personnel tracking and identification due to its low cost, all-weather operation, and high privacy.

[0003] When existing millimeter-wave radar scans a target, it creates a point track around the target. A point track refers to the target coordinate signal output by the radar through internal data processing during the scanning process. The point track is the most basic data unit for achieving target tracking and target trajectory fitting.

[0004] Pedestrian target detection and tracking has always been a research hotspot and challenge in the field of radar technology. Pedestrian targets have small RCS and slow movement speeds, resulting in weak and small target echoes with insignificant Doppler effects, making effective target detection difficult. Furthermore, the surrounding environment of moving pedestrian targets is often complex, easily obscured by ground clutter, building clutter, and tree disturbance signals. In summary, pedestrian targets suffer from weak echoes, strong clutter interference, and Doppler frequencies close to clutter frequencies, resulting in excessive and discontinuous interference points in the radar tracking signal, making it difficult to distinguish effective target points and calculate the target's actual trajectory.

[0005] The following problems exist in current millimeter-wave radar-based personnel tracking and identification: Heuristic data association algorithms, such as the Nearest Neighbor (NN) algorithm, are simple in principle, but have a high association error rate in scenarios with dense targets and high false alarm rates; Kalman or Extended Kalman filtering algorithms cannot accurately estimate the state under non-Gaussian and nonlinear conditions; and simple machine learning algorithms, such as the k-Nearest Neighbor (kNN) algorithm or Logistic Regression (LR) algorithm, cannot accurately and robustly identify targets in pedestrian identification.

[0006] Therefore, there is a need for a real-time detection method and device for radar to detect the relationship between targets and their spatial positions. Summary of the Invention

[0007] The purpose of this invention is to propose a real-time detection method and device for the spatial position relationship between radar-detected targets and areas, which can solve the problems of discontinuous trajectory coordinates of security millimeter-wave radar, numerous discrete signals due to radar clutter interference, high time complexity in calculating the position relationship between coordinates and complex polygonal defense zones, and complex judgment of multi-target motion transformation.

[0008] To achieve the above objectives, the present invention provides a real-time detection method for the relationship between a radar target and its spatial location, comprising:

[0009] The radar's detection plane is divided into a grid-like spatial set, with each grid containing the position coordinates of the current grid and the position coordinates of the adjacent grids;

[0010] The radar signal acquired in real time is updated to the corresponding grid according to the coordinate value. The information of the grid includes: the number of signal trajectory points in the grid, the timestamp of the signal trajectory point record in the grid, and the grid activity value.

[0011] The spatial set is superimposed with the segmented time frame set to form three-dimensional data;

[0012] Extract the feature signal set and convert it into a continuous coordinate curve.

[0013] In an optional approach, obtaining the continuous coordinate curve includes:

[0014] Traverse the data of adjacent grids in the current grid and associate the active grids to form a continuous signal trajectory line;

[0015] A three-dimensional coordinate system is formed by combining the horizontal coordinate system of the continuous trajectory with the time axis; the set of associated signal coordinate points within each trajectory line is projected onto the time axis and the horizontal coordinate plane, and spatially and temporally continuous coordinate points are selected based on the condition that the time projection line is continuous, and the signal is fitted to form the continuous coordinate curve.

[0016] In an optional approach, the method for forming the spatial set includes:

[0017] The radar's detection plane is divided into a grid. The grid is traversed to obtain the position coordinates of the current grid. The position coordinates of the adjacent grids of the current grid are calculated and saved to the current grid.

[0018] In the optional scheme, the radar's detection plane is divided into a preset number of grids according to a preset resolution.

[0019] In an optional approach, the method for determining the active grid includes:

[0020] The activity value of the grid is calculated according to the formula. When the activity value of the grid is greater than the activity threshold, the grid is determined to be an active grid.

[0021] In the optional solution, the formula is as follows:

[0022] v = p*n1 – (t0 – t1)*n2 + s*n3

[0023] Where v is the activity value of the current grid, p is the number of signals in the current grid, n1 is the signal quantity weight; t0 is the current timestamp, t1 is the timestamp of the latest signal in the current grid, n2 is the time weight; s is the number of adjacent active grids, and n3 is the adjacent grid quantity weight.

[0024] In an optional embodiment, forming a continuous signal trajectory line further includes: selecting signals that conform to the movement characteristics of the detected target as signals constituting the trajectory line.

[0025] In an optional embodiment, the movement features include: movement speed and movement direction.

[0026] This invention also discloses a real-time detection device for the relationship between a radar target and its spatial position, comprising:

[0027] The grid module is used to divide the radar's detection plane into a spatial set in the form of a grid. Each grid contains the position coordinates of the current grid and the position coordinates of the adjacent grids.

[0028] The update module is used to update the signals obtained by the radar in real time to the corresponding grid according to the coordinate values. The information of the grid includes: the number of signal trajectory points in the grid, the timestamp of the signal trajectory points in the grid, and the grid activity value.

[0029] The curve module is used to superimpose the spatial set with the segmented time frame set to form three-dimensional data; extract the feature signal set and convert the feature signal set into a continuous coordinate curve.

[0030] In an optional embodiment, the curve module includes: a trajectory line unit and a continuous coordinate curve unit;

[0031] The trajectory line unit is used to traverse the data of adjacent grids of the current grid and associate the active grids therein to form a continuous signal trajectory line;

[0032] The continuous coordinate curve unit is used to form a three-dimensional coordinate system by combining the horizontal coordinate system of the continuous trajectory with the time axis; the set of associated signal coordinate points in each trajectory line is projected onto the time axis and the horizontal coordinate plane, and the spatial and temporal continuous coordinate points are selected based on the condition that the time projection line is continuous, and the signal is fitted to form the continuous coordinate curve.

[0033] The beneficial effects of this invention are as follows:

[0034] This invention can accurately determine the real target and its movement trajectory.

[0035] The system of the present invention has other features and advantages that will be apparent from or will be set forth in detail in the accompanying drawings and following detailed description, which together serve to explain the particular principles of the invention. Attached Figure Description

[0036] The above and other objects, features and advantages of the present invention will become more apparent from the more detailed description of exemplary embodiments of the invention in conjunction with the accompanying drawings.

[0037] Figure 1 A schematic diagram of horizontal plane segmentation and graph data conversion according to an embodiment of the present invention is shown.

[0038] Figure 2 The illustration shows a situation where a signal falls into the grid over a period of time according to an embodiment of the present invention.

[0039] Figure 3 The present invention illustrates a horizontal coordinate system of a filtered continuous trajectory and a three-dimensional coordinate system composed of the horizontal coordinate system of the filtered continuous trajectory and the time axis, according to an embodiment of the present invention. Detailed Implementation

[0040] The invention will now be described in more detail. While preferred embodiments are provided, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the invention will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

[0041] This invention segments a dataset of coordinate points over a period of time into time slices, and spatially divides it into several regions based on the radar detection range horizontal plane (the fineness or resolution of these regions can be adjusted according to different scenarios). Based on the characteristics of common moving targets (people, animals, motor vehicles, etc.), i.e., targets only move in a planar space, specific signals are filtered according to the target's speed and direction of movement for use by the lower-level computer system. To improve computational efficiency, the invention reduces the spatial dimension of the judgment by projecting the target onto the horizontal plane while ignoring changes in target height; it performs real-time calculations based on the relationship between the target's real-time coordinates and the region; it performs parallel calculations of the continuous trajectory of historical targets; and it improves computational efficiency through coarse-grained partitioning and binarization of the planar space and bitwise operations.

[0042] Example 1

[0043] refer to Figure 1 An embodiment of the present invention provides a real-time detection method for the relationship between a radar target and its spatial location, comprising:

[0044] The radar's detection plane is divided into a grid-like spatial set, with each grid containing the position coordinates of the current grid and the position coordinates of the adjacent grids;

[0045] The radar signal acquired in real time is updated to the corresponding grid according to the coordinate value. The information of the grid includes: the number of signal trajectory points in the grid, the timestamp of the signal trajectory point record in the grid, and the grid activity value.

[0046] The spatial set is superimposed with the segmented time frame set to form three-dimensional data;

[0047] Extract the feature signal set and convert it into a continuous coordinate curve.

[0048] Specifically, the detection plane is initialized by dividing the radar's detection plane into a preset number (size) of grids according to a preset resolution. All grids are traversed to obtain the position coordinates of the current grid, calculate the position coordinates of adjacent grids, and save them to the current grid. Each grid contains the position coordinates of the current grid and the position coordinates of its adjacent grids. (Refer to...) Figure 1 It transforms a continuous planar space (grid structure) into a graph data structure, which facilitates the calculation of the longest path (i.e. the true trajectory of the target) and path planning. Figure 1 In the diagram, the nine-square grid on the left represents a grid, indicating that within a certain period of time, there is a signal in grids A1, A2, A6, A9, and A8, but no signal in grids A3, A4, A5, and A7. Figure 1 The right side shows a graph data structure, which can represent multiple possible paths, such as A1-A2-A6-A9-A8, or A1-A2-A6-A8-A9.

[0049] Spatial domain determination and signal coordinate association determination: The signals obtained by the radar in real time are updated to the corresponding grid according to the coordinate values. The grid status is updated (number of signal trajectory points in the grid, timestamp of the signal trajectory points in the grid, grid activity value). The data of adjacent grids of the current grid are traversed (the maximum number of adjacent grids in this embodiment is 8), and the active grids are associated to form a continuous signal trajectory line. The horizontal coordinate system of the continuous trajectory is combined with the time axis to form a three-dimensional coordinate system. The set of associated signal coordinate points in each trajectory line is projected onto the time axis and the horizontal coordinate plane. Based on the continuity of the time projection line, spatially and temporally continuous coordinate points are selected, and the signal is fitted to form the continuous coordinate curve.

[0050] The method for determining whether a grid is active is as follows: Calculate the grid's activity value using a formula. If the activity value of a grid is greater than an activity threshold, the grid is considered active. The activity value of a grid is calculated using the following formula:

[0051] v = p*n1 – (t0 – t1)*n2 + s*n3

[0052] Where v is the activity value of the current grid, p is the number of signals in the current grid, n1 is the signal quantity weight; t0 is the current timestamp, t1 is the timestamp of the latest signal in the current grid, n2 is the time weight; s is the number of adjacent active grids, and n3 is the adjacent grid quantity weight.

[0053] For each frame of signal period, the relevant grid activity value is calculated, and the active grids are associated with the graph data to calculate the longest path. If the grid activity value is less than the activity threshold, the grid is marked as inactive, and the node for calculating the longest path is removed.

[0054] Reference Figure 2 and Figure 3 Trajectory line judgment.

[0055] Figure 2 This represents the situation where the signal falls into the grid over a period of time. Figure 3 The image shows the horizontal coordinate system of the filtered continuous trajectory (left side) and the three-dimensional coordinate system formed by the horizontal coordinate system of the filtered continuous trajectory and the time axis (right side). Figure 2 As can be seen, L1 and L3 form continuous trajectories (identified as real targets), while L2 is noise (which needs to be filtered out).

[0056] Figure 3 In the example of L1, the set of associated signal coordinate points within each trajectory line is projected onto the time axis and the horizontal coordinate plane. By determining whether the time projection line is continuous or not, the points are segmented and filtered to fit the moving trajectory curve with continuous spatial and temporal coordinate points.

[0057] This embodiment has the following beneficial effects:

[0058] (1) The planar space is divided into grids according to resolution and converted into finite element calculation. The coordinate values ​​in a single grid are merged and calculated to achieve large-scale coordinate point sampling.

[0059] (2) The trajectory points have built-in time attributes. The time frame is determined by calculating the timestamp generated by the trajectory points and the time of the current frame to determine whether the time frame contains the trajectory points. The continuous grids are sorted by signal timestamps and the grid activity value is calculated to realize the grid directed graph data structure, which facilitates the calculation of the longest path and the selection of effective feature continuous trajectory signals.

[0060] (3) By projecting the signal within a single continuous trajectory line onto the time axis, a secondary judgment is made on multiple trajectory intersections, single trajectory reversals, etc.

[0061] (4) Based on the detected target movement characteristics (movement speed, movement direction, and variance of signal coordinate values), signals that meet the characteristics can be filtered again to further improve the recognition accuracy.

[0062] (5) Extract feature signals, calculate the most probable path by accumulating time frames, and predict the trajectory coordinates of the next frame by real-time vector calculation based on historical frame coordinates. By iteratively calculating the current time frame data and historical time frame data, real-time analysis of the signal can be achieved, while the computation time complexity of each frame data is basically the same.

[0063] Example 2

[0064] This embodiment provides a real-time detection device for the relationship between a radar target and its spatial location, including:

[0065] The grid module is used to divide the radar's detection plane into a spatial set in the form of a grid. Each grid contains the position coordinates of the current grid and the position coordinates of the adjacent grids.

[0066] The update module is used to update the signals obtained by the radar in real time to the corresponding grid according to the coordinate values. The information of the grid includes: the number of signal trajectory points in the grid, the timestamp of the signal trajectory points in the grid, and the grid activity value.

[0067] The curve module is used to superimpose the spatial set with the segmented time frame set to form three-dimensional data; extract the feature signal set and convert the feature signal set into a continuous coordinate curve.

[0068] In this embodiment, the curve module includes: a trajectory line unit and a continuous coordinate curve unit;

[0069] The trajectory line unit is used to traverse the data of adjacent grids of the current grid and associate the active grids to form a continuous signal trajectory line; the continuous coordinate curve unit is used to form a three-dimensional coordinate system by combining the horizontal coordinate system of the continuous trajectory with the time axis; the set of associated signal coordinate points in each trajectory line is projected onto the time axis and the horizontal coordinate plane, and the spatial and temporal continuous coordinate points are selected based on the condition that the time projection line is continuous, and the signal is fitted to form the continuous coordinate curve.

[0070] The various embodiments of the present invention have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments.

Claims

1. A real-time detection method for the relationship between a radar target and its spatial position, characterized in that, include: The radar's detection plane is divided into a grid-like spatial set, with each grid containing the position coordinates of the current grid and the position coordinates of the adjacent grids; The radar signal acquired in real time is updated to the corresponding grid according to the coordinate value. The information of the grid includes: the number of signal trajectory points in the grid, the timestamp of the signal trajectory point record in the grid, and the grid activity value. The spatial set is superimposed with the segmented time frame set to form three-dimensional data; Extract the feature signal set and convert the feature signal set into a continuous coordinate curve; Obtaining the continuous coordinate curve includes: Traverse the data of adjacent grids in the current grid and associate the active grids to form a continuous signal trajectory line; A three-dimensional coordinate system is formed by combining the horizontal coordinate system of the continuous trajectory with the time axis; the set of associated signal coordinate points within each trajectory line is projected onto the time axis and the horizontal coordinate plane, and spatially and temporally continuous coordinate points are selected based on the condition that the time projection line is continuous, and the signal is fitted to form the continuous coordinate curve. The method for determining the active grid includes: The activity value of the grid is calculated according to the formula. When the activity value of the grid is greater than the activity threshold, the grid is determined to be an active grid. The formula is as follows: v=p*n1–(t0-t1)*n2+s*n3 Where v is the activity value of the current grid, p is the number of signals in the current grid, n1 is the signal quantity weight; t0 is the current timestamp, t1 is the timestamp of the latest signal in the current grid, n2 is the time weight; s is the number of adjacent active grids, and n3 is the adjacent grid quantity weight.

2. The real-time detection method for the relationship between radar-detected targets and their spatial positions according to claim 1, characterized in that, The methods for forming the spatial set include: The radar's detection plane is divided into a grid. The grid is traversed to obtain the position coordinates of the current grid. The position coordinates of the adjacent grids of the current grid are calculated and saved to the current grid.

3. The real-time detection method for the relationship between radar-detected targets and their spatial positions according to claim 1, characterized in that, The radar's detection plane is divided into a preset number of grids according to a preset resolution.

4. The real-time detection method for the relationship between radar-detected targets and their spatial positions according to claim 1, characterized in that, Forming a continuous signal trajectory line also includes: based on the movement characteristics of the detected target, selecting signals that conform to the movement characteristics as signals constituting the trajectory line.

5. The real-time detection method for the relationship between radar-detected targets and their spatial positions according to claim 4, characterized in that, The movement characteristics include: movement speed and movement direction.

6. A real-time detection device for the relationship between a radar target and its spatial position, characterized in that, include: The grid module is used to divide the radar's detection plane into a grid-like spatial set, where each grid contains the position coordinates of the current grid and the position coordinates of the adjacent grids; The update module is used to update the signals obtained by the radar in real time to the corresponding grid according to the coordinate values. The information of the grid includes: the number of signal trajectory points in the grid, the timestamp of the signal trajectory points in the grid, and the grid activity value. The curve module is used to superimpose the spatial set with the segmented time frame set to form three-dimensional data; extract the feature signal set and convert the feature signal set into a continuous coordinate curve; The curve module includes: a trajectory line unit and a continuous coordinate curve unit; The trajectory line unit is used to traverse the data of adjacent grids of the current grid and associate the active grids therein to form a continuous signal trajectory line; The continuous coordinate curve unit is used to form a three-dimensional coordinate system by combining the horizontal coordinate system of the continuous trajectory with the time axis; the set of associated signal coordinate points in each trajectory line is projected onto the time axis and the horizontal coordinate plane, and spatially and temporally continuous coordinate points are selected based on the condition that the time projection line is continuous, and the signal is fitted to form the continuous coordinate curve. The method for determining the active grid includes: The activity value of the grid is calculated according to the formula. When the activity value of the grid is greater than the activity threshold, the grid is determined to be an active grid. The formula is as follows: v=p*n1–(t0-t1)*n2+s*n3 Where v is the activity value of the current grid, p is the number of signals in the current grid, n1 is the signal quantity weight; t0 is the current timestamp, t1 is the timestamp of the latest signal in the current grid, n2 is the time weight; s is the number of adjacent active grids, and n3 is the adjacent grid quantity weight.

Images