Machine gun guided combat system based on vehicle-mounted radar

Abstract

The invention relates to the technical field of radar data analysis, in particular to a machine gun guided combat system based on a vehicle-mounted radar, which comprises a target density detection module, a target detection parameter adjustment module and a non-maximum suppression module, initializing parameters of target detection, performing preliminary target detection on each small region, and counting the number of target detection frames in the preliminary target detection to calculate the target density of each small region; a target detection parameter adjustment module adjusts target detection parameters according to the target density, and performs secondary target detection by using the adjusted parameters to generate a secondary target detection frame set; and the non-maximum suppression module calculates an overlapping threshold according to the target density, and performs non-maximum suppression on the secondary target detection frame set by using the calculated overlapping threshold so as to remove redundant target detection frames.

Description

Technical Field

[0001] The present invention relates to the technical field of radar data analysis, and in particular to a machine gun guided combat system based on vehicle-mounted radar. Background Art

[0002] Traditional object detection systems face multiple technical problems when dealing with complex scenes, the most notable of which include the limitations of fixed parameters and overlapping thresholds.

[0003] First, traditional systems usually use fixed detection parameters, which are difficult to achieve optimal results in detecting objects of different densities. In high-density object areas, fixed parameters may cause multiple objects to be mistakenly merged, thereby increasing the missed detection rate; while in low-density object areas, fixed parameters may cause a single object to be mistakenly segmented into multiple detection boxes, increasing the false detection rate.

[0004] Secondly, the non-maximum suppression module of the traditional system uses a fixed overlap threshold to remove redundant detection frames, which is also unstable in scenes with different densities. In high-density areas, the fixed overlap threshold often mistakenly removes many real target detection frames, resulting in missed detections; while in low-density areas, the fixed overlap threshold may retain too many redundant detection frames, resulting in false detections. These problems not only affect the accuracy and robustness of detection, but also limit the system's ability to be applied in diverse scenes.

[0005] In addition, fixed parameters and overlapping thresholds are also difficult to achieve consistent detection results when facing scenes with different background complexity, further exacerbating the problems of false detection and missed detection. Therefore, in practical applications, traditional target detection systems often require manual intervention to adjust parameters, which not only increases the maintenance cost of the system, but also reduces the real-time performance and reliability of the system. Summary of the invention

[0006] The purpose of the present invention is to provide a machine gun guided combat system based on vehicle-mounted radar to solve the problems raised in the above-mentioned background technology.

[0007] To achieve the above object, the present invention provides the following technical solution: a machine gun guided combat system based on vehicle-mounted radar, comprising a target density detection module, a target detection parameter adjustment module and a non-maximum suppression module, wherein: The target density detection module divides the area covered by the radar signal into multiple small areas, initializes the parameters of target detection, performs preliminary target detection on each small area, and counts the number of target detection frames in the preliminary target detection to calculate the target density of each small area; The target detection parameter adjustment module adjusts the target detection parameters according to the target density, and uses the adjusted parameters to perform secondary target detection to generate a set of secondary target detection frames; The non-maximum suppression module calculates an overlap threshold according to the target density, and uses the calculated overlap threshold to perform non-maximum suppression on the secondary target detection frame set to remove redundant target detection frames.

[0008] As a further improvement of the technical solution, the target density detection module includes an initialization unit, and the initialization unit initializes the target detection parameters including the initial detection threshold and the initial anchor box ratio , use the parameters of initialized target detection to perform preliminary target detection on each small area.

[0009] As a further improvement of the technical solution, the process of the initialization unit performing preliminary target detection specifically includes: Divide the area covered by the radar signal into multiple small areas and initialize the parameters for target detection; Select the detection algorithm based on convolutional neural network as the target detection algorithm, input the radar signal data into the target detection algorithm, and use the target detection algorithm to detect the initial anchor frame according to the initial anchor frame ratio. Generate multiple candidate boxes and calculate the confidence of the target existence for each candidate box, using the initial detection threshold Filter multiple candidate boxes and retain those with confidence exceeding the initial detection threshold The detection frame retained in each small area is used as the preliminary target detection frame; Traverse all small areas, count the number of preliminary target detection frames in each small area, determine whether the center point of each detection frame is located in the small area, if so, add one to the count; and record the number of preliminary target detection frames in each small area.

[0010] As a further improvement of the technical solution, the small area includes a regular rectangular grid and an irregular polygonal grid.

[0011] As a further improvement of the technical solution, the target density detection module includes a target density calculation unit, and the process of the target density calculation unit calculating the target density of each small area specifically includes: Calculate the area of ​​each small region , including the area of ​​rectangular grids and the area of ​​polygonal grids; According to the number of preliminary target detection boxes in each small area And the corresponding small area , calculate the target density of each small area , the calculation formula is .

[0012] As a further improvement of the technical solution, the process of generating a secondary target detection frame by the target detection parameter adjustment module specifically includes: , ,in is the adjusted initial detection threshold; is the adjusted initial anchor box ratio; is the adjustment parameter of the initial detection threshold; is the adjustment parameter of the initial anchor box ratio; Use the adjusted initial detection threshold and the initial anchor box ratio , perform secondary target detection on the radar signal data, use the target detection algorithm to generate a secondary target detection frame, perform secondary target detection on each small area, and integrate them into a set as a secondary target detection frame set.

[0013] As a further improvement of the technical solution, the non-maximum suppression module includes an overlap threshold calculation unit, and the process of calculating the overlap threshold by the overlap threshold calculation unit specifically includes: Get the density baseline value , and according to the target density Calculate the overlap threshold using the following formula: ,in represents the adjusted overlap threshold; is the overlap threshold, indicating a preset initial overlap threshold; is the calculation parameter of the overlap threshold.

[0014] As a further improvement of the technical solution, the non-maximum suppression module includes a non-maximum suppression unit, and the process of the non-maximum suppression unit performing non-maximum suppression specifically includes: Sort each target detection box in the secondary target detection box set from high to low according to the confidence level to generate a sorted list; Initialize an empty list , used to store the final selected target detection box; Take each target detection box from the sorted detection box list in turn , and Add to empty list Calculation and The intersection-and-union ratio of all selected target detection boxes in ; if the intersection-and-union ratio of a target detection box is greater than the calculated adjusted overlap threshold , it will be removed from the sorted list; until all target detection boxes in the sorted list are processed.

[0015] As a further improvement of the technical solution, the area of ​​the rectangular grid is The calculation formula is as follows: ,in Represents the width of the rectangular grid, Represents the height of the rectangular grid.

[0016] As a further improvement of this technical solution, the area of ​​the polygonal mesh The calculation formula is as follows: ,in Represents the vertex coordinates of the polygon, is the number of target vertices in the polygon mesh.

[0017] Compared with the prior art, the present invention has the following beneficial effects: The vehicle-mounted radar-based machine gun guided combat system divides the area covered by the radar signal into multiple small areas, performs preliminary target detection on each small area, and calculates the target density; dynamically adjusts the classification threshold and regression parameters according to these target densities; in high-density areas, relaxes the regression parameters to avoid multiple targets being mistakenly merged and reduce the missed detection rate; in low-density areas, tightens the regression parameters to reduce background noise and mis-segmented detection frames, and reduces the false detection rate. It can adaptively optimize parameters and improve detection accuracy and robustness; In addition, a dynamic overlap threshold is calculated based on the target density. In high-density areas, a lower overlap threshold is set to retain more detection frames, avoid the real targets from being mistakenly removed, and reduce missed detections. In low-density areas, a higher overlap threshold is set to effectively remove redundant detection frames and reduce false detections. These dynamically adjusted overlap thresholds are used for non-maximum suppression to ensure that good detection effects can be maintained in scenes of different densities, further improving the accuracy and reliability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic diagram of the overall module of the present invention; Figure 2 It is a schematic diagram of a target density detection module unit of the present invention; Figure 3 It is a schematic diagram of the non-maximum suppression module unit of the present invention.

[0019] In the figure: 100, target density detection module; 101, initialization unit; 102, target density calculation unit; 200, target detection parameter adjustment module; 300, non-maximum suppression module; 301, overlapping threshold calculation unit; 302, non-maximum suppression unit. DETAILED DESCRIPTION

[0020] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0021] Next, see Figure 1-Figure 3 The present invention provides a technical solution: a machine gun guided combat system based on a vehicle-mounted radar, comprising a target density detection module 100, a target detection parameter adjustment module 200 and a non-maximum suppression module 300.

[0022] The initialization unit 101 in the target density detection module 100 divides the area covered by the radar signal into multiple small areas, initializes the parameters of target detection, performs preliminary target detection on each small area, and counts the number of target detection frames in the preliminary target detection, specifically including: Divide the area covered by the radar signal into a plurality of small areas, wherein the small areas include regular rectangular grids and irregular polygonal grids; Initialize target detection parameters, including initial detection threshold and the initial anchor box ratio , use the parameters of initialized target detection to perform preliminary target detection on each small area; Select the detection algorithm based on convolutional neural network as the target detection algorithm, input the radar signal data into the target detection algorithm, and use the target detection algorithm to detect the initial anchor frame according to the initial anchor frame ratio. Generate multiple candidate boxes and calculate the confidence of the target existence for each candidate box, using the initial detection threshold Filter multiple candidate boxes and retain those with confidence exceeding the initial detection threshold The detection frame retained in each small area is used as the preliminary target detection frame; All small areas are traversed, and the number of preliminary target detection frames in each small area is counted. The specific method is to determine whether the center point of each detection frame is located in the small area. If so, the count is increased by one; the number of preliminary target detection frames in each small area is recorded for subsequent target density calculation.

[0023] The initialization unit 101 divides the area covered by the radar signal into multiple small areas and performs preliminary target detection. The initialization unit 101 can handle scenes of different densities more carefully. The setting of the initial detection threshold and the anchor frame ratio can quickly filter out a large number of low-confidence candidate frames, reducing the burden of subsequent calculations. At the same time, the number of preliminary target detection frames in each small area is counted, which provides basic data for subsequent target density calculations, ensuring that the system can accurately locate high-density and low-density areas.

[0024] The process of calculating the target density of each small area by the target density calculation unit 102 in the target density detection module 100 specifically includes: Calculate the area of ​​each small region , including the area of ​​the rectangular grid and the area of ​​the polygonal grid, where the area of ​​the rectangular grid The calculation formula is as follows: ,in Represents the width of the rectangular grid, Represents the height of the rectangular grid; Area of ​​polygon mesh The calculation formula is as follows: ,in Represents the vertex coordinates of the polygon, is the number of target vertices in the polygonal mesh; According to the number of preliminary target detection boxes in each small area And the corresponding small area , calculate the target density of each small area , the calculation formula is .

[0025] The target density calculation unit 102 enables the system to adaptively adjust subsequent processing parameters through a dynamic density assessment method, thereby improving detection performance and robustness in different density scenarios; in high-density areas, the system can better handle the aggregation of multiple targets; in low-density areas, the system can better identify individual targets and reduce false detections and missed detections.

[0026] The target detection parameter adjustment module 200 adjusts the target detection parameters according to the target density, and uses the adjusted parameters to perform secondary target detection to generate a secondary target detection frame, which specifically includes: According to the target density of all small areas , calculate the average value and use it as the density reference value ; and based on the density reference value And the target density of each small area , adjust the target density to adjust the parameters of target detection, the adjustment formula is as follows: , ,in is the adjusted initial detection threshold; is the adjusted initial anchor box ratio; is the adjustment parameter of the initial detection threshold; is the adjustment parameter of the initial anchor box ratio; Use the adjusted initial detection threshold and the initial anchor box ratio , perform secondary target detection on the radar signal data, use the target detection algorithm to generate a secondary target detection frame, perform secondary target detection on each small area, and integrate them into a set as a secondary target detection frame set.

[0027] The target detection parameter adjustment module 200 calculates the average target density of all small areas as a benchmark value, and adjusts the parameters according to the specific target density of each small area. The system can achieve the best detection effect in different density areas. In high-density areas, the regression parameters are relaxed to reduce the risk of targets being mistakenly merged. In low-density areas, the regression parameters are tightened to reduce background noise and mis-segmented detection frames. This method of adaptively adjusting parameters significantly improves the accuracy and robustness of the system and reduces the need for manual intervention.

[0028] The overlap threshold calculation unit 301 in the non-maximum suppression module 300 calculates the overlap threshold according to the target density, specifically including: Get the density baseline value , and according to the target density Calculate the overlap threshold using the following formula: ,in represents the adjusted overlap threshold; is the overlap threshold, indicating a preset initial overlap threshold; is the calculation parameter of the overlap threshold.

[0029] The overlapping threshold calculation unit 301 dynamically adjusts the overlapping threshold of non-maximum suppression according to the target density; by obtaining the density reference value and calculating the adjusted overlapping threshold according to the target density of each small area, the system can retain more real target detection frames in high-density areas and reduce missed detections; and effectively remove redundant detection frames in low-density areas to reduce false detections.

[0030] The non-maximum suppression unit 302 in the non-maximum suppression module 300 uses the calculated overlap threshold to perform non-maximum suppression on the secondary target detection frame set to remove redundant target detection frames, specifically including: Sort each target detection box in the secondary target detection box set from high to low according to the confidence level to generate a sorted list; Initialize an empty list , used to store the final selected target detection box; Take each target detection box from the sorted detection box list in turn , and Add to empty list Calculation and The intersection-and-union ratio of all selected target detection boxes in ; if the intersection-and-union ratio of a target detection box is greater than the calculated adjusted overlap threshold , it will be removed from the sorted list; until all target detection boxes in the sorted list are processed.

[0031] The non-maximum suppression unit 302 uses the calculated overlapping threshold to process the set of secondary target detection frames, sorts them by confidence and checks the intersection-union ratio one by one, and finally generates a set with redundant detection frames removed; this method not only improves the real-time performance of the system, but also ensures that the detection results in different density scenarios are more accurate and reliable; in high-density areas, multiple high-confidence target detection frames can be effectively retained; in low-density areas, background noise and false detection frames can be effectively removed; in general, the non-maximum suppression unit 302 improves the detection accuracy while also enhancing the robustness and real-time performance of the system.

[0032] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A machine gun guided combat system based on vehicle-mounted radar, characterized in that: It comprises a target density detection module (100), a target detection parameter adjustment module (200) and a non-maximum suppression module (300), wherein: The target density detection module (100) divides the area covered by the radar signal into a plurality of small areas, initializes target detection parameters, performs preliminary target detection on each small area, and counts the number of target detection frames in the preliminary target detection to calculate the target density of each small area; The target detection parameter adjustment module (200) adjusts the target detection parameters according to the target density, and uses the adjusted parameters to perform secondary target detection to generate a set of secondary target detection frames; The non-maximum suppression module (300) calculates an overlap threshold according to the target density, and uses the calculated overlap threshold to perform non-maximum suppression on the secondary target detection frame set, so as to remove redundant target detection frames.

2. The machine gun guided combat system based on vehicle-mounted radar according to claim 1 is characterized in that: The target density detection module (100) comprises an initialization unit (101), wherein the initialization unit (101) initializes target detection parameters including an initial detection threshold. and the initial anchor box ratio , use the parameters of initialized target detection to perform preliminary target detection on each small area.

3. The machine gun guided combat system based on vehicle-mounted radar according to claim 2 is characterized in that: The process of the initialization unit (101) performing preliminary target detection specifically includes: Divide the area covered by the radar signal into multiple small areas and initialize the parameters for target detection; Select the detection algorithm based on convolutional neural network as the target detection algorithm, input the radar signal data into the target detection algorithm, and use the target detection algorithm to detect the initial anchor frame according to the initial anchor frame ratio. Generate multiple candidate boxes and calculate the confidence of the target existence for each candidate box, using the initial detection threshold Filter multiple candidate boxes and retain those with confidence exceeding the initial detection threshold The detection frame retained in each small area is used as the preliminary target detection frame; Traverse all small areas, count the number of preliminary target detection frames in each small area, determine whether the center point of each detection frame is located in the small area, if so, add one to the count; and record the number of preliminary target detection frames in each small area.

4. The machine gun guided combat system based on vehicle-mounted radar according to claim 3 is characterized in that: The small area includes a regular rectangular grid and an irregular polygonal grid.

5. The machine gun guided combat system based on vehicle-mounted radar according to claim 4 is characterized in that: The target density detection module (100) comprises a target density calculation unit (102), and the process of the target density calculation unit (102) calculating the target density of each small area specifically comprises: Calculate the area of ​​each small region , including the area of ​​rectangular grids and the area of ​​polygonal grids; According to the number of preliminary target detection boxes in each small area And the corresponding small area , calculate the target density of each small area , the calculation formula is .

6. The machine gun guided combat system based on vehicle-mounted radar according to claim 5 is characterized in that: The process of generating a secondary target detection frame by the target detection parameter adjustment module (200) specifically includes: , ,in is the adjusted initial detection threshold; is the adjusted initial anchor box ratio; is the adjustment parameter of the initial detection threshold; is the adjustment parameter of the initial anchor box ratio; Use the adjusted initial detection threshold and the initial anchor box ratio , perform secondary target detection on the radar signal data, use the target detection algorithm to generate a secondary target detection frame, perform secondary target detection on each small area, and integrate them into a set as a secondary target detection frame set.

7. The machine gun guided combat system based on vehicle-mounted radar according to claim 6 is characterized in that: The non-maximum suppression module (300) comprises an overlap threshold calculation unit (301), and the process of calculating the overlap threshold by the overlap threshold calculation unit (301) specifically comprises: Get the density baseline value , and according to the target density Calculate the overlap threshold using the following formula: ,in represents the adjusted overlap threshold; is the overlap threshold, indicating a preset initial overlap threshold; is the calculation parameter of the overlap threshold.

8. The machine gun guided combat system based on vehicle-mounted radar according to claim 7 is characterized in that: The non-maximum suppression module (300) comprises a non-maximum suppression unit (302), and the process of the non-maximum suppression unit (302) performing non-maximum suppression specifically comprises: Sort each target detection box in the secondary target detection box set from high to low according to the confidence level to generate a sorted list; Initialize an empty list , used to store the final selected target detection box; Take each target detection box from the sorted detection box list in turn , and Add to empty list Calculation and The intersection-and-union ratio of all selected target detection boxes in ; if the intersection-and-union ratio of a target detection box is greater than the calculated adjusted overlap threshold , it will be removed from the sorted list; until all target detection boxes in the sorted list are processed.

9. The machine gun guided combat system based on vehicle-mounted radar according to claim 5, characterized in that: The area of ​​the rectangular grid The calculation formula is as follows: ,in Represents the width of the rectangular grid, Represents the height of the rectangular grid.

10. The machine gun guided combat system based on vehicle-mounted radar according to claim 5, characterized in that: The area of ​​the polygonal mesh The calculation formula is as follows: ,in Represents the vertex coordinates of the polygon, is the number of target vertices in the polygon mesh.

Images