A Radar Target Tracking Optimization Method Based on Hypothesis Testing

Abstract

This application relates to a radar target tracking optimization method based on hypothesis testing. The method includes: acquiring detection point data of a current target in multiple consecutive frames; estimating the motion direction of the current target based on the detection point data; calculating the mean and variance data of the detection point data in the motion normal direction in multiple frames, and statistically determining the current target within a preset confidence level selection range based on the mean and variance data; determining whether the detection point data in each frame is within the selection range, and if so, determining the detection point data as a real detection point, otherwise determining it as a false detection point. This method makes the target motion trajectory smooth and the speed stable, improving the problems of discontinuous blind zone alarms, false alarms, and missed alarms caused by unstable tracking position and speed of close-range targets by vehicle-mounted radar.

Description

Technical Field

[0001] This application relates to the field of automotive electronics technology, and in particular to a radar target tracking optimization method based on hypothesis testing. Background Technology

[0002] Currently, vehicle-mounted corner radar has an unstable tracking problem for close-range targets. Due to the jumping of the detection point of close-range targets, especially when overtaking from the side, the detection point jumps between the front and rear of the target vehicle, which makes the tracking of the point in the two frames discontinuous and the speed inaccurate, resulting in incorrect judgment of the target's motion state and the formation of false moving targets. Summary of the Invention

[0003] To address the instability issue in the tracking of close-range targets by vehicle-mounted angle radar in existing technologies, this application provides a radar target tracking optimization method based on hypothesis testing.

[0004] A radar target tracking optimization method based on hypothesis testing, the method comprising:

[0005] Acquire detection point data for the current target across multiple consecutive frames;

[0006] Based on the detection point data, the current target's direction of motion is estimated;

[0007] Calculate the mean and variance of the detection point data in the motion normal direction across multiple frames, and based on the mean and variance data, calculate the current target's filtering range value within a preset confidence level.

[0008] Determine whether the detection point data in each frame is within the filtering range. If it is, the detection point data is determined to be a real detection point; otherwise, it is determined to be a false detection point.

[0009] Optionally, acquiring detection point data for multiple consecutive frames of the current target includes:

[0010] Real-time acquisition of detection point data for the current target for 10 to 30 consecutive frames.

[0011] Optionally, the detection point data includes one or more of distance, angle, speed, and signal strength.

[0012] Optionally, estimating the current target's direction of motion based on the detection point data includes:

[0013] Based on the positional changes of the detection point data in multiple frames, the motion direction of the current target is estimated.

[0014] Optionally, estimating the motion direction of the current target based on the position changes of the detection point data across multiple frames includes:

[0015] Calculate the distance r between detection points in every two frames, and decompose the distance into x = r*sin(θ) and y = r*cos(θ) directions in the radar coordinate system;

[0016] The position changes in the X and Y directions are dx and dy, respectively, and based on dx and dy, the mean value in the X direction of the multi-frame detection point data is calculated as u. dx The mean value in the Y direction is u dy Then the direction of motion is head = atan(u dx ,u dy ).

[0017] Optionally, calculating the mean and variance of the detection point data in the motion normal direction across multiple frames includes:

[0018] Calculate the mean distance data u of the detection points in multiple frames respectively. r Variance data var r The average speed data u v Variance data var v .

[0019] Optionally, the filtering range values ​​include distance filtering range values ​​and speed filtering range values, wherein the step of calculating the filtering range values ​​of the current target within a preset confidence level based on mean data and variance data includes:

[0020] Based on the preset reliability, query the normal distribution function table to obtain the adjustment parameter N;

[0021] The distance filtering range value is then... The speed filtering range value is

[0022] Optionally, determining whether the detection point data in each frame is within the filtering range value; if so, the detection point data is a real detection point; otherwise, it is a false detection point; including:

[0023] Determine whether the distance of the detection point data falls within the distance filtering range value;

[0024] If so, determine whether the speed of the detection point data falls within the speed filtering range value; otherwise, determine it as a false detection point.

[0025] If it is determined to be a genuine testing point, it is determined to be a fake testing point.

[0026] Optionally, the preset confidence level is 75% to 95%.

[0027] Optionally, after determining whether the detection point data in each frame is within the filtering range value, and if so, determining the detection point data as a real detection point, otherwise determining it as a false detection point, the method further includes:

[0028] The target motion state is tracked at the real detection points, and the real detection points are judged in real time until the current target leaves the radar detection area.

[0029] The radar target tracking optimization method based on hypothesis testing proposed in this application has the following advantages: This application optimizes the target tracking point information through the statistical results of multi-frame data, making the target movement trajectory smooth and the speed stable, thereby improving the problems of inconsistent blind spot alarms, false alarms, and missed alarms caused by unstable tracking position and speed of vehicle-mounted radar for close-range targets. Attached Figure Description

[0030] Figure 1 Method flow of embodiments of this application Figure 1 .

[0031] Figure 2 This is a system block diagram of an embodiment of this application. Detailed Implementation

[0032] The preferred embodiments of this application will now be described in detail with reference to the accompanying drawings, so that the advantages and features of this application can be more easily understood by those skilled in the art, thereby providing a clearer definition of the scope of protection of this application.

[0033] In the accompanying drawings of the embodiments of this application, the same or similar reference numerals correspond to the same or similar components. In the description of this application, it should be understood that if terms such as "upper", "lower", "left", "right", "top", "bottom", "inner", "outer" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, they are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, the terms used to describe positional relationships in the drawings are only for illustrative purposes and should not be construed as limiting this patent.

[0034] Furthermore, if terms such as "first" or "second" are used for descriptive purposes only, they are mainly used to distinguish different devices, components or parts (the specific types and structures may be the same or different), and are not used to indicate or imply the relative importance or quantity of the indicated devices, components or parts, and should not be construed as indicating or implying relative importance.

[0035] To address the problem in existing technologies where power fluctuations or other interference cause the MCU to fail to read Flash data during the power-on process, resulting in the inability to power on normally, this application provides a method for preventing Flash data read failures, the specific implementation of which is as follows:

[0036] Example 1,

[0037] In such Figure 1-2 In the illustrated embodiment, this application discloses a radar target tracking optimization method based on hypothesis testing. This method includes:

[0038] 100. Acquire detection point data of the current target for multiple consecutive frames; In step 100, this application uses a detection radar to detect the detection point data of the current target for multiple consecutive frames in real time. Acquiring the detection point data of the current target for multiple consecutive frames includes: acquiring the detection point data of the current target for 10 to 30 consecutive frames in real time. The detection point data includes one or more of distance, angle, speed, and signal strength.

[0039] 200. Based on the detection point data, estimate the current target's motion direction. In step 200, estimating the current target's motion direction based on the detection point data includes: estimating the current target's motion direction based on the position changes of multiple frames of detection point data. This estimation includes: calculating the distance r between every two frames of detection points and decomposing the distance into x = r*sin(θ) and y = r*cos(θ) directions in the radar coordinate system; the position changes in the X and Y directions are dx and dy, respectively, and based on dx and dy, the mean value of the X direction of the multiple frames of detection point data is calculated as u. dx The mean value in the Y direction is u dy Then the direction of motion is head = atan(u dx ,u dy ).

[0040] 300. Calculate the mean and variance data of multi-frame detection point data in the motion normal direction, and based on the mean and variance data, calculate the current target's selection range within a preset confidence level. In step 300, calculating the mean and variance data of multi-frame detection point data in the motion normal direction includes: calculating the mean distance data u of the multi-frame detection point data respectively. r Variance data var r The average speed data u v Variance data var vThe filtering range includes distance filtering range and speed filtering range. Specifically, based on mean and variance data, the filtering range for the current target within a preset reliability level is calculated. This includes: querying the normal distribution function table based on the preset reliability level to obtain the adjustment parameter N; thus, the distance filtering range is... Speed ​​filtering range value

[0041] Step 400 involves determining whether the detection point data for each frame falls within the filtering range. If so, the detection point data is considered a real detection point; otherwise, it is considered a false detection point. Step 400 includes determining whether the distance of the detection point data falls within the distance filtering range; if so, determining whether the speed of the detection point data falls within the speed filtering range; otherwise, it is considered a false detection point.

[0042] In this embodiment, the method can be applied to automotive electronic products. The automotive electronic products acquire multi-frame detection point data of targets around the vehicle through onboard radar, process the multi-frame detection point data to filter out real detection points and false detection points. The target tracking point information is optimized through the statistical results of multi-frame data, resulting in smoother target trajectory and more stable speed. This improves the problems of inconsistent blind spot alarms, false alarms, and missed alarms caused by unstable tracking position and speed of close-range targets by onboard radar.

[0043] In some embodiments, acquiring detection point data of the current target across multiple consecutive frames includes: acquiring detection point data of the current target across 10 to 30 consecutive frames in real time. In this embodiment, the application continuously acquires multiple frames of detection point data and replaces the detection point data of the previous frame rate in real time; the current target is tracked through continuous detection. In this embodiment, detection point data of 10, 20, or 30 frames can be selected for statistical calculation. Detection point data includes one or more of distance, angle, velocity, and signal strength. In this embodiment, detection point data includes distance, angle, velocity, signal strength, etc., encompassing various parameters of the detection points of the current target.

[0044] In some embodiments, estimating the motion direction of the current target based on detection point data includes: estimating the motion direction of the current target based on the position changes of multiple frames of detection point data. Estimating the motion direction of the current target based on the position changes of multiple frames of detection point data includes: calculating the distance r between every two frames of detection points, and decomposing the distance into x = r*sin(θ) and y = r*cos(θ) directions in the radar coordinate system; the position changes in the X and Y directions are dx and dy, respectively, and based on dx and dy, calculating the mean value u in the X direction of the multiple frames of detection point data. dx The mean value in the Y direction is u dy Then the direction of motion is head = atan(u dx ,u dy In this embodiment, this application estimates the motion direction of the current target by measuring the position changes of detection points across multiple frames. This can be achieved by first calculating the distance change (dr) between detection point data from every two frames relative to the radar position, and then averaging the distance changes between multiple adjacent detection point data. If dr is close to 0, the current target is considered stationary; otherwise, it is considered moving.

[0045] In some embodiments, calculating the mean and variance data of multi-frame detection point data in the motion normal direction includes: calculating the mean distance data u of the multi-frame detection point data respectively. r Variance data var r The average speed data u v Variance data var v The filtering range includes distance filtering range and speed filtering range. Specifically, based on mean and variance data, the filtering range of the current target within a preset reliability level is calculated, including: querying the normal distribution function table based on the preset reliability level to obtain the adjustment parameter N; thus, the distance filtering range is... Speed ​​filtering range value In this embodiment, before performing calculations, it is assumed that the detection point data of the current target follows a normal distribution in the motion normal direction. The mean and variance of the detection point data across multiple frames are statistically analyzed based on the normal distribution. The motion normal direction is perpendicular to the velocity direction. In this embodiment, since each detection point conforms to a normal distribution, the adjustment parameter value can be found using a function table of the normal distribution based on a preset confidence level. In another embodiment, this application can filter the detection point data based on a preset confidence level of 75%-95%. When the preset confidence level is 95%, the adjustment parameter N is set to 1.96. That is, the distance to the filtering range is... Speed ​​filtering range value

[0046] In some embodiments, determining whether the detection point data of each frame is within a filtering range value, if so, the detection point data is a real detection point; otherwise, it is a false detection point; includes: determining whether the distance of the detection point data falls within a distance filtering range value; if so, determining whether the velocity of the detection point data falls within a velocity filtering range value; otherwise, it is determined to be a false detection point; if so, it is determined to be a real detection point; otherwise, it is determined to be a false detection point. In this embodiment, it can first determine whether the distance of the detection point data falls within a distance filtering range value, and then determine whether the velocity of the detection point data falls within a velocity filtering range; if both conditions are met, it is determined to be a real detection point; otherwise, it is determined to be a false detection point.

[0047] In some embodiments, after determining whether the detection point data of each frame is within the filtering range, if so, the detection point data is determined to be a real detection point; otherwise, it is determined to be a false detection point. The method further includes: tracking the target motion state of the real detection points and determining the real detection points in real time until the current target leaves the radar detection area. In this embodiment, after determining the real detection point of the current target, this application determines the subsequent detection point data of the current target in real time to track the current target until the current target leaves the radar detection area. False detection points are filtered.

[0048] Example 2,

[0049] In such Figure 2 In the illustrated embodiment, this application also provides a radar target tracking optimization method based on hypothesis testing; the specific implementation method is as follows:

[0050] Step 1: First, collect detection point information for the same target for 10 to 30 consecutive frames, including distance r, angle θ, velocity v, signal strength a, etc.

[0051] Step 2: Estimate the target's direction of motion based on the position changes of N frames;

[0052] Step 3: Assuming that the detection points on the same target follow a normal distribution in the direction of their motion velocity, calculate the mean and variance of the position and velocity distribution of these points in this direction;

[0053] Step 4: Select the detection points that meet the confidence level of 75% to 95% as the real detection points of the target, and the others as false detection points or points with low confidence. Use the detection points with high confidence as the real target tracking points.

[0054] Step 5: Use the selected tracking points to track the target's motion state.

[0055] Step 6: Repeat steps 1 through 4 until the target leaves the radar detection area.

[0056] In one specific implementation of the above embodiment, if the radar detects a stationary vehicle A, it collects 10 consecutive frames of detection points (dots(r,θ,v,amp)) on the target vehicle. First, it filters out points with a signal-to-noise ratio below 20 dB. Then, it calculates the position change dr of each two frames of detection points relative to the radar, calculates the average, and if dr is close to 0, it considers the target to be stationary, and records the average position change u. r and variance var r Simultaneously, the mean value of the detection speed u is statistically analyzed. v and variance var v Detection points with a confidence level of 95% or higher were selected based on the mean and variance; that is, points located within the specified range. Speed ​​within range The detection points within the range are considered to be the true target detection points of stationary vehicles. The centroid of these points is used for target tracking. Points outside the range are considered to be false detection points or points with inaccurate detection.

[0057] In another embodiment of the above example, if the radar detects a moving vehicle B, it collects 10 consecutive frames of detection points (dots(r,θ,v,amp)) on the target vehicle. First, it filters out points with a signal-to-noise ratio below 20dB. For a moving target, r is decomposed into two directions in the radar coordinate system: x = r*sin(θ) and y = r*cos(θ). Then, it calculates the positional changes of the points relative to the radar for every two frames: dx and dy, and calculates the mean u. dx u dy Estimate its direction of motion based on position change: head = atan(u dx ,u dy ).

[0058] Calculate the mean distance u from the point to the target in the direction of motion normal. r and variance var r Filter distance meets the range The detection points within the area; secondly, assuming the target moves at a constant speed over a period of time, the velocity u in its direction of motion is statistically analyzed. v and variance var v Filter to meet the range The detection points within the range are considered to be the true target detection points of stationary vehicles. The centroid of these points is used for target tracking. Points outside the range are considered to be false detection points or points with inaccurate detection.

[0059] Example 3,

[0060] This application also discloses a vehicle-mounted terminal, which is connected to a detection radar and runs a program that executes a radar target tracking optimization method based on hypothesis testing as described in Embodiment 1 or Embodiment 2. This application optimizes target tracking point information through statistical results of multi-frame data, resulting in a smooth target trajectory and stable speed. This improves the problems of discontinuous blind spot alarms, false alarms, and missed alarms caused by unstable tracking position and speed of close-range targets by vehicle-mounted radar.

[0061] The embodiments of this application have been described in detail above with reference to the accompanying drawings. However, this application is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of this application.

Claims

1. A radar target tracking optimization method based on hypothesis testing, characterized in that, The method includes: Acquire detection point data for the current target across multiple consecutive frames; Based on the detection point data, the current target's direction of motion is estimated; Calculate the mean and variance of the detection point data in the motion normal direction across multiple frames, and based on the mean and variance data, calculate the current target's filtering range value within a preset confidence level. Determine whether the detection point data in each frame is within the filtering range. If it is, the detection point data is determined to be a real detection point; otherwise, it is determined to be a false detection point. Specifically, this includes determining whether the distance of the detection point data falls within the distance filtering range; if so, determining whether the speed of the detection point data falls within the speed filtering range; otherwise, it is determined to be a false detection point. Specifically, estimating the motion direction of the current target based on the detection point data includes: calculating the distance r between every two frames of detection points, and decomposing the distance into x = r*sin(θ) and y = r*cos(θ) directions in the radar coordinate system; the position changes in the X and Y directions are dx and dy, respectively, and based on dx and dy, the mean value in the X direction of the multi-frame detection point data is calculated. The mean in the Y direction is Then the direction of motion is head = atan( , ); Calculating the mean and variance of the detection point data across multiple frames in the motion normal direction includes: calculating the mean distance data of the detection point data across multiple frames. Variance data Average speed data Variance data The filtering range includes a distance filtering range and a speed filtering range. The step of calculating the current target's filtering range within a preset reliability level based on mean and variance data includes: querying a normal distribution function table based on the preset reliability level to obtain the adjustment parameter N; then the distance filtering range is... The speed filtering range value is .

2. The radar target tracking optimization method based on hypothesis testing according to claim 1, characterized in that, The acquisition of detection point data for multiple consecutive frames of the current target includes: Real-time acquisition of detection point data for the current target for 10 to 30 consecutive frames.

3. The radar target tracking optimization method based on hypothesis testing according to claim 1, characterized in that, The detection point data includes one or more of the following: distance, angle, speed, and signal strength.

4. The radar target tracking optimization method based on hypothesis testing according to claim 1, characterized in that, The preset confidence level is 75% to 95%.

5. The radar target tracking optimization method based on hypothesis testing according to claim 1, characterized in that, After determining whether the detection point data of each frame is within the filtering range value, and if so, the detection point data is determined to be a real detection point; otherwise, it is determined to be a false detection point, the method further includes: The target motion state is tracked at the real detection points, and the real detection points are judged in real time until the current target leaves the radar detection area.

Images