A method and system for synchronous tracking of multi-terminal target positions

Abstract

This invention relates to the field of target detection and tracking technology, specifically a multi-terminal target position synchronization tracking method and system. The invention is used to acquire initial target information from a control console, perform target search based on this initial information, and preprocess the raw data acquired by radar to obtain standard data; perform target detection based on real-time target information acquired by radar, combining adaptive threshold segmentation with morphological reconstruction for weak target enhancement detection; perform target tracking based on the target detection results, introducing an improved nonlinear motion prediction method for tracking complex moving targets; establish a probabilistic fusion model of prediction and detection results to achieve a collaborative optimization mechanism for prediction compensation and target enhancement; and perform multi-terminal synchronization, where multi-terminal synchronization refers to time synchronization, data synchronization, and state synchronization across multiple terminals.

Description

Technical Field

[0001] This invention relates to the field of target detection and tracking technology, specifically a method and system for synchronous tracking of multi-terminal target positions. Background Technology

[0002] Multi-terminal collaborative target tracking has wide applications in video surveillance, unmanned systems, and other fields. However, in practical applications, challenges such as strong nonlinearity of target motion and poor imaging quality are often encountered, making accurate tracking difficult. Existing tracking methods mainly suffer from the following problems: a single linear prediction model is difficult to accurately describe complex motion; traditional target detection algorithms show significant performance degradation under low signal-to-noise ratio conditions; and there is a lack of effective prediction and detection fusion mechanisms. Furthermore, in dynamic environments, rapid response is required to adapt to changing conditions. Summary of the Invention

[0003] (1) Technical problems to be solved

[0004] The purpose of this invention is to provide a method and system for synchronous tracking of multi-target positions, which uses a combination of adaptive threshold segmentation and morphological reconstruction to detect weak targets; employs a multi-model fusion prediction method to achieve target prediction and tracking under complex motion conditions; and realizes a collaborative optimization mechanism for prediction compensation and target enhancement.

[0005] (2) Technical solution

[0006] To achieve the above objectives, in one aspect, the present invention provides a method for synchronous tracking of multi-terminal target positions, the method comprising:

[0007] S10, acquire initial target information sent by the control console, the initial target information including the target's initial position, velocity, and direction; perform target search based on the initial target information, the target search being the radar system acquiring the target's position information in real time after the command vehicle radar and the commander radar receive the target information; perform data preprocessing on the raw data acquired by the radar, the data preprocessing including filtering, data fusion, and coordinate transformation; the data fusion is the fusion of data from multiple radars.

[0008] S20, target detection is performed based on real-time target information acquired by radar. Target detection refers to identifying targets from radar data and extracting target features. Weak target enhancement detection is performed by combining adaptive threshold segmentation with morphological reconstruction. Weak targets refer to target information with low signal-to-noise ratio due to environmental factors.

[0009] S30, target tracking is performed based on the target detection results. The target tracking combines the target's motion model and historical data to predict the target's next position and updates the target's state in real time. An improved nonlinear motion prediction method is introduced to track complex moving targets. A probabilistic fusion model of prediction results and detection results is established to realize a collaborative optimization mechanism of prediction compensation and target enhancement.

[0010] S40, perform multi-terminal synchronization, which refers to time synchronization, data synchronization, and status synchronization among multiple terminals; time synchronization means that the time reference of each radar is consistent; data synchronization means that the target position information acquired by each radar is fused to generate consistent target position information; status synchronization means that the target status information of each radar is synchronized.

[0011] Furthermore, the method for preprocessing the raw data acquired by the radar includes:

[0012] A combined spatial-temporal filtering method is used to filter and suppress noise and enhance the target signal of the original radar signal. High-frequency target signal and low-frequency background noise are separated by spatial small filtering, and static clutter is suppressed and moving target echo is enhanced by temporal multi-frame accumulation. The spatial small filtering is to perform discrete wavelet transform on a single frame radar echo signal to retain high-frequency detail components. The temporal multi-frame accumulation is to perform differential analysis on multiple consecutive frames of signal to eliminate static background.

[0013] The covariance-weighted fusion algorithm is used to fuse target data acquired from multiple radar observations of the same target; the data fusion formula is as follows:

[0014]

[0015] Where D1 = [x1, y1] T And D2 = [x2, y2] T These are the measurements of the same target taken by two radars; P1 and P2 are the covariance matrices of D1 and D2, respectively.

[0016] Furthermore, the method for weak target detection includes:

[0017] Based on the target information, local statistical features are extracted, and the local statistical features of each pixel (x, y) are calculated:

[0018]

[0019] Where I(x,y) represents the gray value of the target at point (x,y); W is the size of the local window; m(x,y) is the mean of the local region centered at point (x,y); and s(x,y) is the standard deviation of the local region.

[0020] The adaptive threshold calculation based on local feature data is performed as follows:

[0021]

[0022] Where T(x,y) is the final determined adaptive threshold; g(x,y) is the gray value of the current processing point; k is the contrast adjustment coefficient; and ∈ is a small positive number to prevent the denominator from being zero.

[0023] Threshold segmentation is performed based on the calculated adaptive threshold. When I(x,y)>T(x,y), the value at each pixel (x,y) is calculated to be 1; otherwise, it is 0.

[0024] Furthermore, the method for weak target enhancement detection that combines adaptive threshold segmentation with morphological reconstruction includes:

[0025] The target image after applying the adaptive threshold segmentation method is used as the initial labeled image, and a mask image is set. The initial labeled image is then iteratively processed. The iterative processing process is as follows: in each iteration, the current labeled image is dilated, and the dilated image and the mask image are subjected to point-by-point minimum operation to obtain a new labeled image. The new labeled image and the previous image are subjected to point-by-point maximum operation to update the labeled image. The iteration is stopped if the difference between the current image and the previous image after each iteration is checked and the difference is less than a preset threshold.

[0026] The reconstructed target image data is obtained, the local contrast features of each pixel in the image are analyzed, and the reconstructed image is weighted using enhancement coefficients to generate an enhanced image.

[0027] Furthermore, the method for tracking the complex moving target includes:

[0028] To predict the motion state of complex moving targets, different motion models are combined. A multi-model fusion prediction method is adopted, which includes uniform motion, uniformly accelerated motion, and turning motion. The formula is as follows:

[0029]

[0030] Where, x k The state of the target at time k, including position and velocity information; z 1:k This is the sequence of observations from time 1 to k; Let p(x) be the state of the i-th motion model at time k; k |z 1:k Given an observation sequence z at time k 1:k At that time, state x k The posterior probability; For a given model When, the observed value zk The likelihood function; Given the model of the previous time step At that time, the current model The transition probability.

[0031] Furthermore, the method for establishing a probabilistic fusion model of prediction results and detection results to achieve a collaborative optimization mechanism for prediction compensation and target enhancement includes:

[0032] The prediction results and detection results are weighted and fused to construct a probabilistic fusion model. Through the collaborative work of prediction compensation and target enhancement, the final target information is generated, achieving a stable tracking effect on the target.

[0033] Based on the same inventive concept, this invention also provides a multi-terminal target location synchronous tracking system, the system comprising: a first data acquisition module, a second target detection module, a third target tracking module, and a multi-terminal synchronization module, wherein the modules are connected in sequence.

[0034] The first data acquisition module is used to acquire the initial target information sent by the control console, perform target search based on the initial target information, and preprocess the raw data acquired by the radar to obtain standardized data.

[0035] The second target detection module is used to perform target detection based on real-time target information acquired by radar. It combines adaptive threshold segmentation with morphological reconstruction to enhance the detection of weak targets, thereby achieving target detection capability under complex motion conditions.

[0036] The third target tracking module is used to track targets based on target detection results, and introduces an improved nonlinear motion prediction method to track complex moving targets; it establishes a probabilistic fusion model of prediction results and detection results to realize a collaborative optimization mechanism of prediction compensation and target enhancement.

[0037] The multi-terminal synchronization module is used to perform multi-terminal synchronization, which refers to time synchronization, data synchronization, and status synchronization among multiple terminals. Time synchronization means that the time reference of each radar is consistent. Data synchronization means that the target position information acquired by each radar is fused to generate consistent target position information. Status synchronization means that the target status information of each radar is synchronized.

[0038] (3) Beneficial effects

[0039] Compared with the prior art, the beneficial effects of the present invention are: it realizes enhanced detection of targets with low signal-to-noise ratio, uses a multi-model fusion prediction method to predict and track targets under complex motion conditions, and runs a collaborative optimization mechanism of prediction compensation and target enhancement. Attached Figure Description

[0040] Figure 1 This is a flowchart of a multi-terminal target position synchronous tracking method according to Embodiment 1 of the present invention;

[0041] Figure 2 This is a schematic diagram of the module composition of a multi-terminal target position synchronous tracking system according to Embodiment 2 of the present invention. Detailed Implementation

[0042] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0043] Example 1: As Figure 1 As shown in the figure, this embodiment provides a method for synchronous tracking of multi-terminal target positions, the method including:

[0044] S10, acquire initial target information sent by the control console, the initial target information including the target's initial position, velocity, and direction; perform target search based on the initial target information, the target search being the radar system acquiring the target's position information in real time after the command vehicle radar and the commander radar receive the target information; perform data preprocessing on the raw data acquired by the radar, the data preprocessing including filtering, data fusion, and coordinate transformation; the data fusion is the fusion of data from multiple radars.

[0045] S20, target detection is performed based on real-time target information acquired by radar. Target detection refers to identifying targets from radar data and extracting target features. Weak target enhancement detection is performed by combining adaptive threshold segmentation with morphological reconstruction. Weak targets refer to target information with low signal-to-noise ratio due to environmental factors.

[0046] S30, target tracking is performed based on the target detection results. The target tracking combines the target's motion model and historical data to predict the target's next position and updates the target's state in real time. An improved nonlinear motion prediction method is introduced to track complex moving targets. A probabilistic fusion model of prediction results and detection results is established to realize a collaborative optimization mechanism of prediction compensation and target enhancement.

[0047] S40, perform multi-terminal synchronization, which refers to time synchronization, data synchronization, and status synchronization among multiple terminals; time synchronization means that the time reference of each radar is consistent; data synchronization means that the target position information acquired by each radar is fused to generate consistent target position information; status synchronization means that the target status information of each radar is synchronized.

[0048] For example, on January 15, 2024, a multi-target synchronous tracking experiment was conducted at a military base training ground. At 08:35:00, the control console issued initial target drone information: position N41°24'15", E119°27'32", altitude 312 meters, ground speed 168 km / h, heading 275 degrees. The main radar and cooperative radar acquired the target for the first time at 08:35:12 and 08:35:15 respectively, measuring distances of 7823 meters and 7819 meters. After data preprocessing, the target fusion positioning error was 3.8 meters, and the velocity measurement error was 2.3 km / h. During the target enhancement detection phase, the original signal-to-noise ratio (SNR) of the main radar was 8.2 dB, and that of the cooperative radar was 7.8 dB. After activating the adaptive threshold, the SNR increased to 12.5 dB and 11.9 dB respectively. Target tracking prediction showed that it was turning at a rate of 12 degrees / second, with an altitude change of -5 meters / second. Through multi-terminal synchronization, the clock deviation between the two radars is controlled within 0.002 seconds, and the position error after fusion is 2.8 meters, and the speed error is 1.5 kilometers per hour.

[0049] It should be noted that the method for preprocessing the raw data acquired by the radar includes:

[0050] S11 employs a combined spatial-temporal filtering method to filter and suppress noise in the original radar signal while enhancing the target signal. A small spatial filter separates the high-frequency target signal from the low-frequency background noise, and a multi-frame accumulation in the temporal domain suppresses static clutter and enhances the moving target echo. The small spatial filter performs a discrete wavelet transform on a single-frame radar echo signal to preserve high-frequency detail components. The multi-frame accumulation in the temporal domain performs differential analysis on multiple consecutive frames to eliminate static background.

[0051] S12, a covariance-weighted fusion algorithm is used to fuse target data acquired from multiple radar observations of the same target; the data fusion formula is:

[0052]

[0053] Where D1 = [x1, y1] T And D2 = [x2, y2] T These are the measurements of the same target taken by two radars; P1 and P2 are the covariance matrices of D1 and D2, respectively.

[0054] For example, based on the original target signal, the target echo intensity is -75dBm, ground clutter is -72dBm, and meteorological clutter is -70dBm. After processing with a 7×7 pixel filter window, the target intensity is reduced to -73dBm, and the clutter is reduced to -85dBm. After performing 8 frames of time-domain accumulation, a clutter suppression ratio of 15dB and a target enhancement ratio of 6dB are obtained. The target distances measured by the two radars are 7925 meters and 7928 meters, with azimuth angles of 35 degrees and 82 degrees, and measurement errors of 4.2 meters and 3.8 meters, respectively. After data fusion, the target distance is determined to be 7926 meters, the fusion error is reduced to 2.1 meters, and the confidence level reaches 95.8%.

[0055] It should be noted that the method for weak target detection includes:

[0056] S21, Based on the target information, perform local statistical feature extraction and calculate the local statistical features of each pixel (x, y):

[0057]

[0058]

[0059] Where I(x,y) represents the gray value of the target at point (x,y); W is the size of the local window, which is usually an odd number (such as 7, 9, 11); m(x,y) is the mean of the local region centered at point (x,y); and s(x,y) is the standard deviation of the local region, which is used to characterize the dispersion of the gray value distribution within the region.

[0060] S22, Adaptive threshold calculation based on local feature data, the method is as follows:

[0061]

[0062] Where T(x,y) is the final determined adaptive threshold; g(x,y) is the gray value of the current processing point; k is the contrast adjustment coefficient, with a value range of [0.2,0.5]; ∈ is a small positive number to prevent the denominator from being zero, usually taken as 0.001.

[0063] S23. Perform threshold segmentation based on the calculated adaptive threshold. When I(x,y)>T(x,y), calculate that the value at each pixel (x,y) is 1; otherwise, it is 0.

[0064] For example, statistical features were extracted from the target region: the center point grayscale value was 125, the region mean was 118, and the standard deviation was 12.5; the background region mean was 85, the standard deviation was 8.3, and the signal-to-noise ratio was 6.8 dB. A contrast coefficient of 0.35 was set, and an adaptive threshold was calculated using a 9×9 window. The threshold for the target region was set at 110, and the threshold for the background region was set at 95. The final segmentation accuracy reached 92.5%.

[0065] It should be noted that the method for weak target enhancement detection that combines adaptive threshold segmentation with morphological reconstruction includes:

[0066] S231, the target image after applying the adaptive threshold segmentation method is used as the initial labeled image, and a mask image is set; the initial labeled image is iteratively processed, the iterative processing process is as follows: in each iteration, the current labeled image is dilated, the dilated image and the mask image are subjected to point-by-point minimum operation to obtain a new labeled image, the new labeled image and the previous image are subjected to point-by-point maximum operation to update the labeled image; by checking the difference between the current image and the previous image after each iteration, if the difference is less than a preset threshold, the iteration is stopped.

[0067] S232: Obtain the reconstructed target image data, analyze the local contrast features of each pixel in the image, and use enhancement coefficients to weight the reconstructed image to generate an enhanced image.

[0068] For example: the initial marked image shows 225 target pixels and an initial outline sharpness of 65%; it is reconstructed using a 5×5 structuring element with 256 gray levels and an 88% boundary preservation rate. After 8 iterations, the target area stabilizes at 320 pixels from the initial 225 pixels, the outline sharpness is improved to 92%, and the final iteration difference is reduced to 1.8%.

[0069] It should be noted that the method for tracking the complex moving target includes:

[0070] S31 combines different motion models to predict the motion state of complex moving targets, employing a multi-model fusion prediction method that includes uniform motion, uniformly accelerated motion, and turning motion. The formula is as follows:

[0071]

[0072] Where, x k The state of the target at time k, including position and velocity information; z 1:k This is the sequence of observations from time 1 to k; Let p(x) be the state of the i-th motion model at time k; k |z 1:k Given an observation sequence z at time k 1:k At that time, state x k The posterior probability; For a given model When, the observed value z k The likelihood function; Given the model of the previous time step At that time, the current model The transition probability.

[0073] For example, the prediction results using multiple models are as follows: the uniform motion model predicts a position error of 45 meters, with a model weight of 0.25 and a likelihood value of 0.68; the uniformly accelerated motion model predicts a position error of 38 meters, with a weight of 0.35 and a likelihood value of 0.75; and the turning motion model predicts a position error of 22 meters, with a weight of 0.40 and a likelihood value of 0.82. After final fusion calculation, the target position is determined to be 41°24'19"N, 119°27'22"E, with the overall error controlled within 18 meters.

[0074] It should be noted that the method for establishing a probabilistic fusion model of prediction results and detection results to achieve a collaborative optimization mechanism for prediction compensation and target enhancement includes:

[0075] S311 weightedly fuses the prediction results and detection results to construct a probabilistic fusion model. Through the collaborative work of prediction compensation and target enhancement, the final target information is generated, achieving a stable tracking effect on the target.

[0076] For example, by setting the prediction result weight to 0.45 and the detection result weight to 0.55 for probability fusion, the optimized tracking error was reduced to 12 meters, and tracking stability was improved by 25%. Throughout the tracking process, the average position error was 15 meters, the velocity measurement error was 1.5 meters per second, and the heading measurement error was 1.2 degrees. The data refresh rate was maintained at 10 Hz, the processing latency was 85 ms, and the tracking stability reached 98.5%. In complex electromagnetic environments, 25 dB clutter suppression was achieved, the minimum detectable signal-to-noise ratio was 6 dB, and the effective tracking distance was extended to 12 kilometers.

[0077] Example 2: Based on the same inventive concept, such as Figure 2 As shown in the figure, this embodiment also provides a multi-terminal target location synchronization tracking system, the system including: a first data acquisition module, a second target detection module, a third target tracking module, and a multi-terminal synchronization module, the modules being connected in sequence.

[0078] The first data acquisition module is used to acquire the initial target information sent by the control console, perform target search based on the initial target information, and preprocess the raw data acquired by the radar to obtain standardized data.

[0079] The second target detection module is used to perform target detection based on real-time target information acquired by radar. It combines adaptive threshold segmentation with morphological reconstruction to enhance the detection of weak targets, thereby achieving target detection capability under complex motion conditions.

[0080] The third target tracking module is used to track targets based on target detection results, and introduces an improved nonlinear motion prediction method to track complex moving targets; it establishes a probabilistic fusion model of prediction results and detection results to realize a collaborative optimization mechanism of prediction compensation and target enhancement.

[0081] The multi-terminal synchronization module is used to perform multi-terminal synchronization, which refers to time synchronization, data synchronization, and status synchronization among multiple terminals. Time synchronization means that the time reference of each radar is consistent. Data synchronization means that the target position information acquired by each radar is fused to generate consistent target position information. Status synchronization means that the target status information of each radar is synchronized.

[0082] It should be noted that the specific methods by which each module performs operations in the system described in the above embodiments have been described in detail in the embodiments related to the method, and will not be elaborated here.

[0083] Finally, it should be noted that although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. 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 multi-terminal target position synchronization tracking method, characterized by, The method includes the following steps: S10, acquire initial target information sent by the control console, including the target's initial position, velocity, and direction; perform target search based on the initial target information, wherein the target search is performed by the radar system acquiring the target's position information in real time after the command vehicle radar and the commander radar receive the target information; perform data preprocessing on the raw data acquired by the radar, including filtering, data fusion, and coordinate transformation; the data fusion is the fusion of data from multiple radars; S20, target detection is performed based on real-time target information acquired by radar. Target detection refers to identifying targets from radar data and extracting target features. Weak target enhancement detection is performed by combining adaptive threshold segmentation with morphological reconstruction. Weak targets refer to target information with low signal-to-noise ratio due to environmental factors. S30, target tracking is performed based on the target detection results. The target tracking combines the target's motion model and historical data to predict the target's next position and updates the target's state in real time. An improved nonlinear motion prediction method is introduced to track complex moving targets. A probabilistic fusion model of prediction results and detection results is established to realize a collaborative optimization mechanism of prediction compensation and target enhancement. S40, perform multi-terminal synchronization, which refers to time synchronization, data synchronization, and status synchronization among multiple terminals; time synchronization means that the time reference of each radar is consistent; data synchronization means that the target position information acquired by each radar is fused to generate consistent target position information; status synchronization means that the target status information of each radar is synchronized. The method for weak target detection includes: According to the target information, local statistical feature extraction is performed, and local statistical features of each pixel point are calculated : ； ； wherein, represents the gray value of the target at point ; is the local window size; is the local region mean centered at point ; is the local region standard deviation; The adaptive threshold calculation based on local feature data is performed as follows: ； in, The final adaptive threshold; This represents the grayscale value of the current processing point. This is the contrast adjustment coefficient; To prevent small positive numbers with a denominator of zero; Threshold segmentation is performed based on the calculated adaptive threshold. > When, then calculate each pixel point The value is 1 if it is selected, otherwise it is 0; The method for weak target enhancement detection that combines adaptive threshold segmentation with morphological reconstruction includes: The target image after applying the adaptive threshold segmentation method is used as the initial labeled image, and a mask image is set. The initial labeled image is then iteratively processed. The iterative processing process is as follows: in each iteration, the current labeled image is dilated, and the dilated image and the mask image are subjected to point-by-point minimum operation to obtain a new labeled image. The new labeled image and the previous image are subjected to point-by-point maximum operation to update the labeled image. The iteration is stopped if the difference between the current image and the previous image after each iteration is checked and the difference is less than a preset threshold. The reconstructed target image data is obtained, the local contrast features of each pixel in the image are analyzed, and the reconstructed image is weighted using enhancement coefficients to generate an enhanced image.

2. The multi-terminal target position synchronization tracking method as described in claim 1, characterized in that, The method for preprocessing the raw data acquired by the radar includes: A combined spatial-temporal filtering method is used to filter and suppress noise and enhance the target signal from the original radar signal. Spatial wavelet filtering separates the high-frequency target signal from the low-frequency background noise, while temporal multi-frame accumulation suppresses static clutter and enhances the moving target echo. The spatial wavelet filtering performs discrete wavelet transform on a single-frame radar echo signal to preserve high-frequency detail components. The temporal multi-frame accumulation performs differential analysis on multiple consecutive frames to eliminate static background. The covariance-weighted fusion algorithm is used to fuse target data acquired from multiple radar observations of the same target; the data fusion formula is as follows: ； in, and These are measurements taken by two radars of the same target; and They are and The covariance matrix.

3. The multi-terminal target position synchronization tracking method as described in claim 1, characterized in that, The methods for tracking complex moving targets include: To predict the motion state of complex moving targets, different motion models are combined. A multi-model fusion prediction method is adopted, which includes uniform motion, uniformly accelerated motion, and turning motion. The formula is as follows: ； ； in, For the goal at all times The status, including position and velocity information; From time 1 to The sequence of observations; For the first A motion model at time... The state; For at any time Given observation sequence At that time, state The posterior probability; For a given model At that time, the observed value The likelihood function; Given the model of the previous time step At that time, the current model The transition probability.

4. The multi-target position synchronization tracking method as described in claim 1, characterized in that, The method for establishing a probabilistic fusion model of prediction results and detection results to achieve a collaborative optimization mechanism for prediction compensation and target enhancement includes: The prediction results and detection results are weighted and fused to construct a probabilistic fusion model. Through the collaborative work of prediction compensation and target enhancement, the final target information is generated, achieving a stable tracking effect on the target.

5. A multi-terminal target position synchronous tracking system, used to perform the method according to any one of claims 1-4, characterized in that, The system includes: a first data acquisition module, a second target detection module, a third target tracking module, and a multi-terminal synchronization module, which are connected in sequence. The first data acquisition module is used to acquire the initial target information sent by the control console, perform target search based on the initial target information, and preprocess the raw data acquired by the radar to obtain standardized data. The second target detection module is used to perform target detection based on real-time target information acquired by radar, and combines adaptive threshold segmentation method with morphological reconstruction to enhance the detection of weak targets, thereby achieving target detection capability under complex motion conditions; The third target tracking module is used to track targets based on target detection results, and introduces an improved nonlinear motion prediction method to track complex moving targets; it establishes a probabilistic fusion model of prediction results and detection results to realize a collaborative optimization mechanism of prediction compensation and target enhancement. The multi-terminal synchronization module is used to perform multi-terminal synchronization, which refers to time synchronization, data synchronization, and status synchronization among multiple terminals. Time synchronization means that the time reference of each radar is consistent. Data synchronization means that the target position information acquired by each radar is fused to generate consistent target position information. Status synchronization means that the target status information of each radar is synchronized. The method for weak target detection includes: Local statistical features are extracted based on the target information, and the values ​​of each pixel are calculated. Local statistical characteristics: ； ； in, Indicates the target is at point The grayscale value at that location; This refers to the size of a local window. For the point The mean of the local region centered on it; This represents the standard deviation for a local region. The adaptive threshold calculation based on local feature data is performed as follows: ； in, The final adaptive threshold; This represents the grayscale value of the current processing point. This is the contrast adjustment coefficient; To prevent small positive numbers with a denominator of zero; Threshold segmentation is performed based on the calculated adaptive threshold. > When, then calculate each pixel point The value is 1 if it is selected, otherwise it is 0; The method for weak target enhancement detection that combines adaptive threshold segmentation with morphological reconstruction includes: The target image after applying the adaptive threshold segmentation method is used as the initial labeled image, and a mask image is set. The initial labeled image is then iteratively processed. The iterative processing process is as follows: in each iteration, the current labeled image is dilated, and the dilated image and the mask image are subjected to point-by-point minimum operation to obtain a new labeled image. The new labeled image and the previous image are subjected to point-by-point maximum operation to update the labeled image. The iteration is stopped if the difference between the current image and the previous image after each iteration is checked and the difference is less than a preset threshold. The reconstructed target image data is obtained, the local contrast features of each pixel in the image are analyzed, and the reconstructed image is weighted using enhancement coefficients to generate an enhanced image.

Images