A method and system for processing tracks based on radar plot generation dynamic clutter map

Abstract

This invention relates to the field of radar dynamic clutter maps, and specifically to a method and system for processing flight tracks based on radar point traces to generate dynamic clutter maps. This invention processes the amplitude and velocity information of nine adjacent cells at the range and azimuth of a target cell to obtain a preliminary judgment of the target cell's clutter situation. The flight track information is then compared with the similarity features of the corresponding clutter map cells in a pre-stored flight track clutter map. If it is a clutter track, it is removed from the radar display; if it is a suspected target, it is uploaded to the radar display interface for display. This invention employs a flight track clutter suppression method based on clutter maps, which can effectively suppress fixed clutter and residual slow-moving clutter, thereby eliminating false alarms caused by the movement of low-speed targets at range and azimuth.

Description

Technical Field

[0001] This invention relates to the field of radar dynamic clutter maps, and in particular to a method and system for processing flight tracks based on generating dynamic clutter maps from radar point traces. Background Technology

[0002] Due to the influence of the operating environment, low-altitude sounding radars are easily alerted by fixed targets (such as billboards, windmills, building air conditioners, etc.) and low-speed small targets (birds, small scatterers, etc.). The operating environment faced by low-altitude sounding radars is quite variable, so fixed clutter maps cannot meet the operational requirements.

[0003] Traditional clutter maps have slow update speeds or high clutter margins. Traditional clutter maps adjust their detection performance using clutter coefficients, and the update speed of these maps is highly unpredictable for detecting slow-moving targets. This is as disclosed in Chinese invention patent "Method, Apparatus and Equipment for Detecting Small, Slow Targets Based on Dynamic Threshold Factor" (Application No.: CN202310121544.2). Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method for processing flight tracks based on dynamic clutter maps generated from radar points. This invention processes the amplitude and velocity information of nine adjacent cells at the target cell's range and azimuth to predict the clutter situation of the target cell. The flight track information is then compared with the track features within the corresponding cells of a pre-stored clutter map. If a clutter track is detected, it is removed from the radar display; if it is a suspected target, it is uploaded to the radar display interface for display. This invention employs a clutter map-based track clutter suppression method, which can effectively suppress fixed clutter and residual slow-moving clutter, thereby eliminating false alarms caused by the movement of low-speed targets at range and azimuth.

[0005] The technical solution of this invention is: a method for generating dynamic clutter maps based on radar spot patterns, characterized by comprising the following steps:

[0006] Step 1: Divide the target detection space into detection units and establish clutter feature data with three features: azimuth, range, and velocity. The clutter features correspond to the background features of different azimuth and range units. The feature library is established as follows: Each input echo information data corresponds to a detection unit. Each detection unit in the detection space stores seven features: average amplitude of the echo signal, average velocity, amplitude density, velocity entropy, peak point velocity and maximum velocity difference, peak point velocity and minimum velocity difference, and fourth-order central moment. These seven features constitute the feature library of a detection unit.

[0007] Step 2: Statistically analyze the clutter map with an update period of n min. The clutter map is analyzed by dividing it into intervals and counting the frequency of different features in each interval of n min, including the frequency distribution of the average amplitude, average velocity, amplitude density, velocity entropy, peak velocity-minimum velocity, peak velocity-maximum velocity, and fourth-order central moment of the same target detection space. The clutter background feature vector is composed by selecting the largest value, forming a feature library for all detection units, thus generating the clutter map.

[0008] Step 3: Update the clutter map information based on the input echo information. For newly input echo points, calculate the average amplitude, average velocity, amplitude density, velocity entropy, peak velocity, maximum / minimum velocity difference, and fourth-order central moment of the echo signals of the points participating in the aggregation. Take the center of the aggregated point as the central unit, and add eight units adjacent to it in terms of distance and orientation to form a nine-square grid. Calculate the similarity of the clutter map feature vectors with the nine neighboring units, using cosine similarity as the metric. The nine neighboring units are selected using the method of selecting the largest neighbor, which means finding the value with the largest similarity in the nine-square grid as the feature similarity of the aggregated point.

[0009] Step 4: The similarity value ranges from [-1, 1], where 1 indicates that the feature is completely similar to the background feature and -1 indicates that the feature is completely dissimilar.

[0010] According to the method for generating dynamic clutter maps based on radar spot patterns as described above, the characteristic feature is that: in the initial state of step 1, the feature data of each detection unit in the detection space is set to 0;

[0011] According to the method for generating dynamic clutter maps based on radar spot patterns as described above, the characteristic is that: in step 1, the amplitude density is calculated as follows: Where A i The amplitude of each point in the detection unit, where N refers to the N points; velocity entropy: pV i For velocity V i The probability of detection; peak velocity: the velocity corresponding to the maximum amplitude in the detection space; minimum velocity: the minimum velocity value in the detection space; maximum velocity: the maximum velocity value in the detection space; fourth-order central moment: SD represents the standard deviation of amplitude.

[0012] According to the method for generating dynamic clutter maps based on radar spot patterns as described above, the characteristic is that n is 10 min.

[0013] According to the method for generating dynamic clutter maps based on radar spot patterns as described above, the characteristic feature is that the formula for calculating the feature cosine similarity between the echo and each detection unit is as follows:

[0014]

[0015] Where C i The seven features of the unit background are D. i These are the seven characteristics of an echo.

[0016] According to the method for generating dynamic clutter maps based on radar spot patterns as described above, the method is characterized in that: in step four, different spot patterns with different similarities are assigned different spot quality. The spot quality is assigned according to the characteristics of the clutter region. The larger the sum of the seven feature frequencies, the stronger the clutter characteristics, and the more clutter region it belongs to. The higher the similarity, the worse the spot quality.

[0017] According to the method for generating dynamic clutter maps based on radar spot patterns as described above, the characteristic is that in step four, 10 spots with poor spot quality are used as the starting batch, and 3 spots with high spot quality are used as the starting batch.

[0018] This invention also discloses a system for processing flight tracks based on generating dynamic clutter maps using radar spot patterns. The system includes a feature calculation module, a clutter storage module, a clutter update module, and a spot quality discrimination module. The feature calculation module calculates the features of each input echo and statistically analyzes the frequency of spot features. The clutter storage module stores and saves clutter feature data from all detected spaces, extracts clutter background feature vectors, and forms a clutter map. The clutter update module periodically updates the clutter feature information based on the input echo information and calculates the similarity between each input echo and the clutter map feature vectors of its nine neighboring units using cosine similarity. The spot quality discrimination module assigns spot quality values ​​based on the clutter region discrimination results and the similarity between spot features and clutter map features. Different batching difficulties can be assigned based on spot quality during track batching.

[0019] The beneficial effects of this invention are: effectively suppressing fixed clutter and slow clutter residue, and eliminating false alarms caused by the movement of low-speed targets in range and azimuth. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating the method of the present invention. Detailed Implementation

[0021] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0022] like Figure 1 As shown, this invention discloses a method for generating dynamic clutter maps based on radar tracks, and a system for processing tracks using dynamic clutter maps. The system includes a feature calculation module, a clutter storage module, a clutter update module, and a track quality discrimination module. The specific steps of this invention are as follows.

[0023] Step 1: Divide the target detection space into detection units and establish clutter feature data with three features: azimuth, range, and velocity. The clutter features correspond to the background features of different azimuth and range units. The clutter feature data is the standard library for target discrimination.

[0024] The feature library is built as follows: each input echo data corresponds to a detection unit, and each detection unit in the detection space stores the average amplitude of the echo signal (mean of the mode of the interval). ), average speed (mean of the mode of the interval) The detection space contains seven features: amplitude density, velocity entropy (characterizing the disorder of velocity), difference between peak velocity and maximum velocity, difference between peak velocity and minimum velocity, and fourth-order central moment. These seven features constitute the feature library of a detection unit. Initially, the feature data of each detection unit in the detection space is set to 0.

[0025] Amplitude density: Where A i The amplitude of each point in the detection unit, where N refers to the number of points.

[0026] Velocity dimensional entropy: pV i For velocity V i probability

[0027] Peak point velocity: The velocity corresponding to the maximum value of the detected spatial amplitude;

[0028] Minimum speed: The minimum speed value detected in the space;

[0029] Maximum speed: The maximum speed value detected in the space;

[0030] The fourth central moment is also called the kurtosis factor: SD represents the standard deviation of amplitude.

[0031] The feature library of all detection units is continuously updated, and the feature library of each detection unit includes the above 7 features.

[0032] Using these seven features can effectively distinguish cluttered cells from clean cells.

[0033] Step 2: The update cycle is set to n min, where n is typically 10 min. Longer cycles result in more stable clutter map statistics. Due to the wide range of feature variations, interval division is generally used. For example, the amplitude interval span is (maximum amplitude - minimum amplitude) / 5, resulting in 5 intervals. Other features are divided similarly. The frequency of each interval for different features within n min is statistically analyzed, including the frequency distribution of the average amplitude, average velocity, amplitude density, velocity entropy, peak velocity - minimum velocity, peak velocity - maximum velocity, and fourth-order central moment. The clutter background feature vector is formed by selecting the largest possible feature value (the median of the interval) corresponding to the maximum frequency of different features, creating a feature library for all detection units, thus generating the clutter map.

[0034] Step 3: Update the clutter map information based on the input echo information. For newly input echo points, calculate the average amplitude, average velocity, amplitude density, velocity entropy, peak velocity, maximum / minimum velocity difference, and fourth-order central moment of the echo signals of the points participating in the aggregation. Using the center of the aggregated point as the central unit, add eight units adjacent in terms of distance and azimuth to form a nine-grid unit. Each of the nine units has its own clutter feature data. Calculate the similarity of the clutter map feature vectors with the nine neighboring units (the nine-grid) using cosine similarity as the metric. For the nine neighboring units, use the largest neighborhood selection method; that is, find the value with the largest similarity in the nine-grid as the feature similarity of the aggregated point.

[0035] The formula for calculating the characteristic cosine similarity between the echo and each detector unit is as follows:

[0036]

[0037] Where C i The seven features of the unit background are D. i Seven characteristics of echo

[0038] Step 4: The similarity value ranges from [-1, 1], where 1 indicates complete similarity to the background features, and -1 indicates complete dissimilarity. Higher similarity indicates a higher probability of clutter. Different similarity values ​​are assigned different point quality to different aggregation points. The point quality assignment can be adjusted according to the characteristics of the clutter region. Generally, the larger the sum of the frequencies of the seven major features, the stronger the clutter characteristics, indicating a heavy clutter region. The point quality in a heavy clutter region should differ from that in a clean region at the same similarity value. Higher similarity results in lower point quality. This increases the difficulty of initial batching. For example, for poor point quality, batches can start with 10 points; for high-quality points, batches can start with 3 points. This effectively shields temporary clutter aircraft.

[0039] 4. The feature calculation module calculates the features and statistical frequency of each input echo information, i.e., it calculates seven features for each input echo information, completing the data calculation and extraction in step 1 and the frequency statistics in step 2. The clutter storage module stores and saves clutter feature data for all probed spaces, extracts clutter background feature vectors, and forms a clutter map. The clutter update module periodically updates the clutter feature information based on the input echo information, and calculates the similarity between each input echo information and the clutter map feature vectors of its nine neighboring units using cosine similarity, completing the calculation in step 3. The spot quality discrimination module assigns spot quality values ​​based on the clutter region discrimination results and the similarity between spot features and clutter map features. During track batching, different batching difficulties can be assigned based on spot quality, completing the calculation in step 4.

[0040] This invention utilizes seven eigenvalues, each representing the maximum frequency of a feature, to improve the detection probability of slow-moving targets, effectively suppress fixed and low-speed targets, and also effectively suppress residual clutter generated by high-speed targets to a certain extent. Clutter is generally caused by scattering from fixed targets, residual clutter, and reflections from the ground and sea surface. Compared to targets, clutter occurs more frequently and its amplitude distribution is more concentrated. This invention fully considers the concentration of clutter signal energy, the concentration of velocity component distribution, and the frequency of clutter signal occurrence. The seven selected features are calculated to be significant. In addition, background clutter is constantly changing with the environment and weather; sliding updates of background features help improve the detection probability. The determination of track quality comprehensively considers the degree of change in the background clutter area and the similarity of features. The difference between the echo and the background is used to differentiate the difficulty of track batching. For clean areas, the target batching difficulty can be set low, while for heavily cluttered areas, the target batching difficulty can be set high, realizing the function of zoned batching. Furthermore, due to the range and azimuth scattering characteristics of radar target detection, a nine-square grid neighborhood is used to fully consider the functional characteristics of the radar itself.

[0041] This invention improves the clutter detection method, which can not only effectively suppress fixed targets and low-speed targets, but also effectively suppress clutter residue generated by high-speed targets to a certain extent.

Claims

1. A method for generating dynamic clutter maps based on radar spot patterns, characterized in that: Includes the following steps: Step 1: Divide the target detection space into detection units and establish clutter feature data with three features: azimuth, range, and velocity. The clutter features correspond to the background features of different azimuth and range units. The feature library is established as follows: Each input echo information data corresponds to a detection unit. Each detection unit in the detection space stores seven features: average amplitude of the echo signal, average velocity, amplitude density, velocity entropy, peak point velocity and maximum velocity difference, peak point velocity and minimum velocity difference, and fourth-order central moment. These seven features constitute the feature library of one detection unit. Step 2: Statistically analyze the clutter map with a 10-minute update cycle. The clutter map is analyzed by dividing it into intervals and calculating the frequency of different features in each interval over 10 minutes. This includes the frequency distribution of the average amplitude, average velocity, amplitude density, velocity entropy, peak velocity-minimum velocity, peak velocity-maximum velocity, and fourth-order central moment of the same target detection space. A clutter background feature vector is constructed using a method that selects the largest values, forming a feature library for all detection units, thus generating the clutter map. Step 3: Update the clutter map information based on the input echo information. For newly input echo points, calculate the average amplitude, average velocity, amplitude density, velocity entropy, peak velocity difference between the maximum and minimum velocities, and fourth-order central moment of the echo signals of the points participating in the aggregation. Take the center of the aggregated point as the central unit, and add eight units adjacent to it in terms of distance and orientation to form a nine-square grid. Calculate the similarity of the clutter map feature vectors with the nine neighboring units, using cosine similarity as the metric. The nine neighboring units are selected using the method of selecting the largest neighbor, which means finding the value with the largest similarity in the nine-square grid as the feature similarity of the aggregated point. Step 4: The similarity value ranges from [-1, 1], where 1 indicates complete similarity to the background features and -1 indicates complete dissimilarity. Different similarity values ​​are assigned different point quality values ​​to different aggregation points. The point quality value is adjusted according to the characteristics of the clutter region. The larger the sum of the frequencies of the 7 features, the stronger the clutter characteristics, and the more cluttered the region. The higher the similarity, the worse the point quality.

2. The method for generating dynamic clutter maps based on radar spot patterns as described in claim 1, characterized in that: In the initial state of step 1, the feature data of each detection unit in the detection space is set to 0.

3. The method for generating dynamic clutter maps based on radar spot patterns as described in claim 1, characterized in that: In step 1, the amplitude density is calculated as follows: A i The amplitude of each point in the detection unit, where N refers to the N points; velocity entropy: pV i For velocity V i The probability of detection; peak velocity: the velocity corresponding to the maximum amplitude in the detection space; minimum velocity: the minimum velocity value in the detection space; maximum velocity: the maximum velocity value in the detection space; fourth-order central moment: SD represents the standard deviation of amplitude. This represents the arithmetic mean of the detection units.

4. The method for generating dynamic clutter maps based on radar spot patterns as described in claim 1, characterized in that: The formula for calculating the feature cosine similarity between the echo and each detection unit is as follows: ； Where C i The seven features of the unit background are D. i These are the seven characteristics of an echo.

5. The method for generating dynamic clutter maps based on radar spot patterns as described in claim 1, characterized in that: In step four, the batch size starts with 10 dots of poor quality and 3 dots of high quality.

6. A system for processing flight tracks based on generating dynamic clutter maps from radar spot patterns, comprising a feature calculation module, a clutter storage module, a clutter update module, and a spot quality discrimination module, characterized in that: The feature calculation module calculates the features and statistical point feature frequency of each input echo information. The clutter storage module is used to store and save the clutter feature data of all detection spaces, extract the clutter background feature vector, and form a clutter map. The clutter update module updates the clutter feature information periodically with the input echo information and calculates the similarity between each input echo information and the clutter map feature vector of the nine neighboring units through cosine similarity. The spot quality discrimination module assigns spot quality values ​​based on the clutter region discrimination results and the similarity between spot features and clutter map features. Different batching difficulties can be assigned based on spot quality when starting a flight path batch.

7. The system for processing flight tracks based on generating dynamic clutter maps using radar spot patterns according to claim 6, characterized in that: The method for generating dynamic clutter maps based on radar spot patterns as described in any one of claims 1 to 5 is adopted.

Images