Auxiliary shooting method and device based on image recognition

Abstract

The present invention belongs to the technical field of digital real-soldier combat systems, and specifically relates to an auxiliary shooting method and device based on image recognition, including: collecting and updating position data, spatial angle data and image data in real time through a digital transmitter; when a weapon is detected to be fired, the currently collected position data, spatial angle data and image data are packaged to form digital ammunition and sent to the guidance and control end; the guidance and control end performs trajectory deduction based on the received position data and spatial angle data to obtain a basic trajectory, and then corrects the basic trajectory according to the image data to identify the striking part and obtain the final trajectory; based on the final trajectory, a collision damage model is determined, and an injury determination result is generated and returned to the soldier terminal. The present invention improves the strike accuracy of the digital transmitter and improves the overall confidence of the digital real-soldier combat system.

Description

Technical Field

[0001] The present invention belongs to the technical field of digital real-life combat systems, and in particular relates to an auxiliary shooting method and device based on image recognition. Background Art

[0002] The laser transmitter of the laser engagement system is mechanically fixed. After calibration, the laser emission angle can basically be guaranteed to coincide with the aiming line. When shooting, the laser beam is less affected by external interference, which can ensure high shooting accuracy. When the laser is irradiated to the laser receiving probe group on the individual equipment, the casualties are judged based on whether the laser probes in different parts receive the laser and bring it into the damage model.

[0003] The digital live-fire combat system uses data streams instead of lasers to simulate bullets, effectively avoiding the harm of high-power lasers to human eyes. It also solves the problem that the linear diffusion of lasers makes it easy to hit distant targets but difficult to hit close-range targets.

[0004] However, in terms of shooting accuracy, the digital transmitter used in the digital combat system is still difficult to compare with the traditional laser transmitter, which is mainly reflected in the case of medium and long-distance combat. The digital transmitter uses various sensors to collect the position and posture of the weapon. Since the accuracy of the sensor is limited, the data will float and drift within a certain range, which will inevitably cause errors. If the sensor angle fluctuates within 0.2°, there will be a position deviation of 0.35 meters at a target of 100 meters, and a position deviation of 1.4 meters at a target of 400 meters. Although the data floating problem can be improved through filtering algorithms or the use of higher-precision sensors, the cost will increase significantly, and the long-distance hit part is still difficult to identify and cannot be brought into the damage model, which is contrary to the original design intention of the digital combat system and is not conducive to the development of the digital combat system. Summary of the invention

[0005] Purpose of the invention: The purpose of the present invention is to address the deficiencies of the prior art and provide an auxiliary shooting method and device based on image recognition, so as to improve the strike accuracy of the digital transmitter and improve the overall confidence of the digital actual combat system.

[0006] Technical solution: The auxiliary shooting method based on image recognition of the present invention comprises the following steps: Collect and update position data, spatial angle data and image data in real time through digital transmitters; When the weapon is detected to be firing, the currently collected position data, spatial angle data and image data are packaged into digital ammunition and sent to the guidance and control end; The guidance and control end performs trajectory deduction based on the received position data and spatial angle data to obtain the basic trajectory, and then corrects the basic trajectory based on the image data to identify the impact site and obtain the final trajectory; The collision damage model is determined based on the final trajectory, and the injury determination result is generated and returned to the soldier terminal.

[0007] To further improve the above technical solution, the angle data sent to the control terminal includes several groups of angle data before firing, each group of angle data includes a pitch angle, a roll angle and a heading angle, and the control terminal processes the several groups of angle data as follows: Normalize several groups of angle data to get: , where Ang is a number of sets of angle data, is the mean, is the standard deviation; Define the optimal fusion set to obtain valid observation data: in, is the observation data, and are preset consistency and distribution thresholds.

[0008] Furthermore, the identification of the striking parts of the collected image data includes: first identifying whether there is a person in the image data, if so, automatically selecting and identifying the person, and then identifying the current human body part, including identifying at least one of the head, neck, trunk and limbs, and when multiple parts are identified at the same time, assigning the parts according to the preset weights; the identified striking parts form a direct aiming identification code.

[0009] Furthermore, identifying the impact position of the collected image data includes the following steps: The input image data Convert to grayscale image ; For grayscale images Perform normalization: ,in, and are the mean and standard deviation of the image respectively; The normalized image Input into the convolutional neural network, after one or more layers of convolution operations, each layer contains convolution kernels, and the convolution kernel size is , the step length is , get the feature map : ,in, is the convolution kernel weight, is the bias, * indicates the convolution operation, is the activation function; The feature map output by the last convolutional layer Flatten it into vector f, input it into the fully connected layer, and get the output vector: ,in, is the weight of the fully connected layer, is bias; The predicted probability of each category is calculated by the Softmax function: , where N is the number of categories, including head, neck, left chest, right chest, left and right abdomen, right abdomen, back and limbs, is the predicted probability of the ith class.

[0010] Furthermore, it also includes: measuring the distance between the weapon and the target, after the weapon is fired, an indirect aiming identification code is formed, the position of the indirect aiming identification code in the image data is adjusted based on the measured distance, and the target part is re-identified based on the adjusted indirect aiming star identification code; the adjustment of the indirect aiming star position is also based on at least one or more of the following parameters: gravity; air resistance; type of weapon; type of ammunition; initial velocity of ammunition firing; when performing ballistic deduction: when the target distance is relatively close, the weight of the attitude sensor data is increased; when the target distance is relatively far, the weight of the image recognition information is increased; at a long distance, the shooting hit judgment is mainly based on the attitude sensor data, and the damage judgment is mainly based on the image recognition information.

[0011] Furthermore, the calculation process of the basic trajectory is as follows: in, is the position vector, is the initial position vector, is the velocity vector, is the initial velocity vector, is the acceleration vector and t is the time.

[0012] Furthermore, the damage determination method process is as follows: A kinetic interceptor is created at the target. The damage of the warhead to the kinetic interceptor is determined by the coverage of the bullet in the area swept by the warhead along the relative velocity direction. The collision point is calculated as follows: in, is the coordinate of the collision point P in the target coordinate system, are the coordinates of discrete points on the kinetic interceptor, is the relative velocity vector component, and t is the time.

[0013] The auxiliary shooting device based on image recognition for implementing the above method comprises: A digital transmitter, the digital transmitter includes a weapon main control, an RTK positioning module, a distance measurement module, a posture sensor, a camera, an image recognition module, a communication module, and a weapon fixture. The RTK positioning module is used to collect high-precision position data, the distance measurement module is used to measure the distance between the weapon and the target, the posture sensor is used to collect spatial angle data of the weapon, and the camera is used to collect image data of the aiming direction; the image recognition module is used to obtain the image data collected by the camera and identify the human body and human body parts in the image data; the weapon fixture is used to fix the digital transmitter to weapons of different calibers; the weapon main control is used to obtain the position data collected by the RTK positioning module, the spatial angle data collected by the posture sensor, and the image data processed by the image recognition module, and form digital ammunition to be sent to the soldier terminal through the communication module; A soldier terminal is connected to the weapon main control and guidance control terminals of the digital transmitter, respectively, and is used to send digital ammunition to the guidance control terminal and obtain the injury determination result generated by the guidance control terminal; The guidance and control terminal performs trajectory deduction based on the position data and spatial angle data to obtain the basic trajectory, and then identifies the impact site according to the image data to correct the basic trajectory to obtain the final trajectory. Based on the final trajectory, the collision damage model is determined to generate an injury determination result.

[0014] Furthermore, the deviation of the position data obtained by repeated positioning and collection by the RTK positioning module does not exceed 10 centimeters. The RTK positioning module adopts an integrated structure of the positioning module and the antenna, or adopts a split structure to integrate the RTK module into the weapon main control and lead out the antenna feeder; the attitude sensor includes a gyroscope, an electronic compass, an inclinometer and a dual-antenna directional module, and the collected spatial angle data includes pitch angle, roll angle and heading angle; the camera is selected from at least one of an ordinary camera, an infrared camera and a thermal imaging camera, and the focal length of the camera is adjustable, including optical zoom and digital zoom.

[0015] Furthermore, it also includes a flash simulator, a flame sensor and a split power supply structure, the flash simulator is used to simulate the flame when the weapon is fired, and the flame sensor is used to identify the fire generated during live ammunition shooting; the split power supply structure includes a host end and a battery that are separately arranged, and the battery is installed through a replaceable battery tube.

[0016] Beneficial effect: Compared with the prior art, the advantages of the present invention are: the present invention provides an auxiliary shooting method based on image recognition, which identifies the target through the image recognition module. Combined with the precise distance information provided by the ranging module, the auxiliary aiming system can calculate the precise ballistic landing point, especially in long-range shooting, by adjusting the position of the aiming star in real time, the hit rate is significantly improved, and the problem that the slight error of the sensor causes the ballistic drift to be too large during long-range shooting, resulting in blurred hit range is solved, and the confidence of the ballistic restoration and troop damage judgment of the parallel battlefield digital twin equipment in the digital actual combat system is improved.

[0017] When the weapon is fired, the digital transmitter packages multiple sets of data such as positioning, spatial angle, and image recognition into digital ammunition, which is transmitted to the soldier terminal and the guidance and control terminal through the communication module. The guidance and control terminal uses this data to perform trajectory deduction and damage judgment, and feeds back the results to the soldier terminal to achieve real-time feedback and adjustment.

[0018] The image recognition module identifies different parts of the human body and combines the data from the posture sensor to make accurate damage judgments. The system can dynamically adjust the weight distribution according to the target distance to ensure the accuracy of judgments at different distances. Especially when shooting at long distances, the weight of image recognition information is increased, which enhances the reliability of damage judgments.

[0019] In the auxiliary aiming mode, the system can automatically identify the target and lock the distance parameters, form an indirect aiming identification code and send it to the guidance and control end for comprehensive processing together with other data. Even when firing continuously, the system can process data efficiently and reduce channel pressure.

[0020] The present invention significantly improves shooting accuracy, combat adaptability and accuracy of damage assessment, enhances the combat effectiveness and survivability of soldiers in various combat environments, and provides the military with an efficient, precise, practical and safe training method. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a digital transmitter with image recognition function; Figure 2 Relationship between the crosshair and distance for image recognition; Figure 3 Model for character recognition.

[0022] In the figure: digital transmitter; 1a, digital transmitter body; 1b, camera; 1c, flash simulator / flame sensor; 1d, ranging module; 1e, weapon fixture; 1f, battery; 1g, RTK positioning module. DETAILED DESCRIPTION

[0023] The technical solution of the present invention is described in detail below with reference to the accompanying drawings, but the protection scope of the present invention is not limited to the embodiments.

[0024] Example 1: Figure 1 The image recognition-based auxiliary shooting device shown in the figure has the following core components: 1a is the main body of the digital transmitter, which contains the weapon main control, attitude sensor, and image recognition module. The weapon main control integrates the communication module and the vibration sound collector. 1b is the camera lens. The camera can be externally mounted and supports manual adjustment of the focal length, including optical zoom and digital zoom. At the same time, the display screen is optional and can be configured as required to adjust the camera focal length. 1c is the flash simulator / flame sensor. The flash simulator is used to simulate the flame when the weapon is fired. The flame sensor receives infrared light and is used to identify the flame generated during live ammunition shooting. 1d is the ranging module, which identifies the distance from the target point to the muzzle. 1e is the weapon clamp, which is used to fix the digital transmitter to the weapon barrel. It should be noted that the clamps for weapons of different calibers are different. 1f is the battery cartridge, which can be adapted to different battery models. The upper cover of the battery cartridge shown in the figure can be removed for battery replacement. 1g is the RTK positioning module, which is used to collect data from positioning satellites and provide high-precision positioning for the digital transmitter.

[0025] The RTK positioning module can adopt an integrated module with positioning module and antenna, or a split type. The RTK positioning module is integrated into the weapon main control, and the antenna feed line is led out for installation. The antenna installation position needs to be prominent to facilitate better signal reception and increase stability.

[0026] There are many ways to power a digital transmitter. Figure 1 The figure shows the method of installing the host and the battery separately, and an integrated power supply can also be used, that is, the battery is installed in the host.

[0027] The attitude sensor is used to collect the spatial angle data of the weapon, including pitch angle, roll angle and heading angle. The attitude sensor includes a gyroscope, an electronic compass, an inclinometer, and a dual-antenna directional module.

[0028] The camera is used to collect images in the direction of the weapon's aiming. For weapons with different effective ranges, lenses with different focal lengths are used. The longer the focal length, the smaller the viewing angle, and the farther the image can be clearly captured. Cameras include ordinary cameras, infrared cameras, and thermal imaging cameras. Ordinary cameras mainly receive visible light, which is close to the image observed by the human eye and is suitable for well-lit conditions. Infrared cameras mainly receive near-infrared light and are suitable for dimly lit conditions. Thermal imaging cameras collect thermal radiation from objects and are less interfered by external visible light, so they are suitable for most situations.

[0029] The digital transmitter can be integrated with a rangefinder to obtain the distance between the weapon and the target, and then pass the distance data to the image recognition module. The distance measurement can be done manually or automatically.

[0030] The image recognition module is used to process images collected by the camera. By training the algorithm model, it mainly identifies humans and different parts of the human body, including the head, neck, torso (left chest, right chest, left and right abdomen, right abdomen, back) and limbs, so as to facilitate the guidance and control to make damage judgments. The image recognition module processes image data as follows: First, the input image is preprocessed. Convert to grayscale image : a +b +c in, They are the red, green and blue channels of the image respectively, and a, b and c are the parameter adjustment ratios.

[0031] right Perform normalization: in, and are the mean and standard deviation of the image respectively.

[0032] Will Through a layer of convolutional neural network, each layer contains convolution kernels, the convolution kernel size is , the step length is , then the feature map of the first layer output It can be expressed as: in, is the convolution kernel weight, is the bias, * indicates the convolution operation, is the activation function.

[0033] The feature map of the first layer Classification is performed through the fully connected layer and flattened to vector f. The output of the fully connected layer can be expressed as: in, is the weight of the fully connected layer, is bias.

[0034] At this point, the probability of recognition for each category is calculated: Where N is the number of categories, here left and right chest, left and right abdomen, back, is the predicted probability of the ith class.

[0035] Here, the category with the highest probability is selected as the final recognition result to obtain the part.

[0036] The overall implementation of the auxiliary shooting scheme based on image recognition is as follows: the digital transmitter should be installed parallel to the barrel and must be bound to the soldier terminal before the exercise. The soldier terminal is connected to the guidance and control end through the communication base station, and the data flow is normal. When the digital transmitter detects that the weapon is firing, the weapon master control packages the current positioning, spatial angle and image recognition information into digital ammunition, which is sent to the guidance and control end through the soldier terminal. These data are simulated by firepower at the guidance and control end to provide data support for the ballistic restoration of digital twin equipment in parallel battlefields and the judgment of troop damage. Since there is a certain fluctuation in the angle data, multiple sets of angle data can be packaged and sent together before firing. The guidance and control end fuses multiple sets of angle data to deduce the basic trajectory, and then corrects the trajectory according to a certain weight based on the image recognition results to form a trajectory closest to the actual shooting, and then forms a collision damage model with the target and makes a judgment on the injury, and finally feeds the processing results back to the soldier terminal.

[0037] Normalization of multiple sets of angle data can be obtained: Among them, Ang is multiple sets of angle data, is the mean, is the standard deviation.

[0038] Here we define the optimal fusion set to obtain valid observation data: in, is the observation data, and are preset consistency and distribution thresholds.

[0039] The basic trajectory is calculated as follows: in, is the position vector, is the velocity vector, is the acceleration vector and t is the time.

[0040] The damage determination method is based on volume overlap analysis. A kinetic interceptor is created at the target. The geometric shapes of the warhead and the kinetic interceptor are simplified, and the warhead can be approximated as a cylinder.

[0041] The damage of the warhead to the kinetic interceptor is determined by the coverage of the bullet by the kinetic interceptor in the area swept by the warhead along the relative velocity direction. The collision point is calculated as follows: in, is the coordinate of the collision point P in the target coordinate system, are the coordinates of discrete points on the kinetic interceptor, is the relative velocity vector component, and t is the time.

[0042] Through the above processing, the warhead collision damage model can be realized and the injury determination can be carried out.

[0043] like Figure 2 The relationship between the image recognition crosshair and the distance shown in the figure: The figure contains two examples of crosshairs. One is the center circle, which is the direct aiming center and is fixed in the center of the image; the other is the crosshair, which is the indirect aiming center and can be moved. In the figure, the distance between the person and the camera gradually increases from 2a to 2d. When the image frame remains unchanged, the displayed person becomes smaller. If there is no distance parameter, regardless of whether the focal length changes, both crosshairs will remain unchanged in the middle of the image, such as Figure 2 As shown in a.

[0044] When the assisted aiming function is turned on, distance measurement will be introduced and combined with character recognition. When the image recognition module recognizes the person in the picture, it will automatically measure the distance. When different parts of the human body are recognized, the image recognition module will lock and store the current distance parameters (the distance parameters will not be updated temporarily unless a new person is recognized again). At this time, the indirect aiming center will calculate the landing point based on the current distance, and the image recognition module will identify the target part again according to the location of the indirect aiming star landing point. Soldiers can also use the indirect aiming star as the basis for aiming. After the weapon is fired, an indirect aiming identification code is formed and transmitted to the guidance and control end together with other data for trajectory and damage deduction.

[0045] After inputting the distance parameter into the image recognition module and turning on the assisted aiming, the image processing module will use the distance information to estimate the trajectory point. The farther the distance, the lower the indirect aiming center point will move. The assisted aiming function of the crosshair can be turned on or off manually, and the calibration point of the rangefinder module coincides with the direct aiming strike point. The input parameters for the change of the indirect aiming center point include distance, gravity, air resistance, weapon type, ammunition type, and ammunition initial velocity, which supports pre-setting and real-time correction.

[0046] like Figure 3As shown, this is a person recognition process for this example. First, identify whether there is a person in the picture. If so, automatically select and identify it, as shown in Figure 3a; then identify the current body parts, including the head, neck, torso and limbs, as shown in Figure 3b. If multiple parts are identified at the same time, the possible body parts are allocated according to certain weights. When the weapon is fired, the identified strike part will form a direct aiming identification code and be uploaded together with other data. The guidance and control background will perform trajectory deduction based on the position, direction, and pitch at the time of shooting, and will perform trajectory deduction and injury judgment again based on the deduction results and identification code according to certain algorithm weights, and finally feedback the processing results to the soldier terminal through the base station.

[0047] The weight ratio is dynamically adjusted according to the shooting distance. When the target is close, the weight of the attitude sensor data is higher; when the target is farther away, the weight of the image recognition information will gradually increase. At long distances, the weight distribution reaches the limit, and the shooting hit judgment is still mainly based on the attitude sensor data, while the damage judgment tends to be mainly based on the image recognition information.

[0048] Regardless of the distance, if differentiated shooting hit judgment results are generated multiple times, that is, the posture sensor data deduction does not hit, but the image recognition has a hit; or the posture sensor data deduction has a hit, but the image recognition does not hit. The guidance and control end system can generate targeted warning information without actively interfering with the shooting, making it easier for security personnel to find out the cause.

[0049] For auxiliary shooting schemes based on image recognition, hit and damage judgment requires learning through a large number of samples to train the optimal algorithm and damage model, so as to correct the trajectory solution error caused by the fluctuation of the attitude sensor output data through image information, increase the accuracy of ballistic solution and the confidence of troop damage judgment.

[0050] After the digital transmitter detects the firing signal and identifies the person model at the same time, the picture of the firing moment can be sent to the control along with the firing data. The picture contains the distance parameter and person identification information. If continuous firing is detected, the number of images sent can be reduced to reduce the pressure on the channel.

[0051] When the image processing module does not recognize a person, the digital transmitter can still shoot normally. At this time, the guidance and control will only rely on the data of the attitude sensor when calculating the trajectory.

[0052] As described above, although the present invention has been shown and described with reference to specific preferred embodiments, it should not be construed as limiting the present invention itself. Various changes in form and details may be made without departing from the spirit and scope of the present invention as defined in the appended claims.

Claims

1. An auxiliary shooting method based on image recognition, characterized in that: The following steps are involved: Collect and update position data, spatial angle data and image data in real time through digital transmitters; When the weapon is detected to be firing, the currently collected position data, spatial angle data and image data are packaged into digital ammunition and sent to the guidance and control end; The guidance and control end performs trajectory deduction based on the received position data and spatial angle data to obtain the basic trajectory, and then corrects the basic trajectory based on the image data to identify the impact site and obtain the final trajectory; The collision damage model is determined based on the final trajectory, and the injury determination result is generated and returned to the soldier terminal.

2. The auxiliary shooting method based on image recognition as claimed in claim 1, characterized in that: The angle data sent to the control terminal include several groups of angle data before firing, each group of angle data includes pitch angle, roll angle and heading angle, and the control terminal processes the several groups of angle data as follows: Normalize several groups of angle data to get: , where Ang is a number of sets of angle data, is the mean, is the standard deviation; Define the optimal fusion set to obtain valid observation data: in, is the observation data, and are preset consistency and distribution thresholds.

3. The auxiliary shooting method based on image recognition as claimed in claim 1, characterized in that: The striking part recognition of the collected image data includes: firstly recognizing whether there is a person in the image data, and if so, automatically selecting and recognizing the person, and then recognizing the current body part, including recognizing at least one of the head, neck, torso and limbs. When multiple parts are recognized at the same time, the parts are allocated according to preset weights; the recognized striking part forms a direct aiming recognition code.

4. The auxiliary shooting method based on image recognition as claimed in claim 1, characterized in that: The identification of the striking position of the collected image data includes the following steps: The input image data Convert to grayscale image ; For grayscale images Perform normalization: ,in, and are the mean and standard deviation of the image respectively; The normalized image Input into the convolutional neural network, after one or more layers of convolution operations, each layer contains convolution kernels, and the convolution kernel size is , the step length is , get the feature map : ,in, is the convolution kernel weight, is the bias, * indicates the convolution operation, is the activation function; The feature map output by the last convolutional layer Flatten it into vector f, input it into the fully connected layer, and get the output vector: ,in, is the weight of the fully connected layer, is bias; The predicted probability of each category is calculated by the Softmax function: , where N is the number of categories, including head, neck, left chest, right chest, left and right abdomen, right abdomen, back and limbs, is the predicted probability of the ith class.

5. The auxiliary shooting method based on image recognition as claimed in claim 4, characterized in that: Also includes: Measuring the distance between the weapon and the target. After the weapon is fired, an indirect aiming identification code is generated. The position of the indirect aiming identification code in the image data is adjusted based on the measured distance. The target part is re-identified based on the adjusted indirect aiming identification code. When conducting trajectory simulation: when the target is close, increase the weight of the attitude sensor data; when the target is far away, increase the weight of the image recognition information; at long distances, the shooting hit judgment is mainly based on the attitude sensor data, and the damage judgment is mainly based on the image recognition information.

6. The auxiliary shooting method based on image recognition as claimed in claim 1, characterized in that: The calculation process of the basic trajectory is as follows: in, is the position vector, is the initial position vector, is the velocity vector, is the initial velocity vector, is the acceleration vector and t is the time.

7. The auxiliary shooting method based on image recognition as claimed in claim 6, characterized in that: The damage determination method process is as follows: A kinetic interceptor is created at the target. The damage of the warhead to the kinetic interceptor is determined by the coverage of the bullet in the area swept by the warhead along the relative velocity direction. The collision point is calculated as follows: in, is the coordinate of the collision point P in the target coordinate system, are the coordinates of discrete points on the kinetic interceptor, is the relative velocity vector component, and t is the time.

8. A device for implementing the auxiliary shooting method based on image recognition as claimed in claim 1, characterized in that: include: A digital transmitter, the digital transmitter includes a weapon main control, an RTK positioning module, a distance measurement module, a posture sensor, a camera, an image recognition module, a communication module, and a weapon fixture. The RTK positioning module is used to collect high-precision position data, the distance measurement module is used to measure the distance between the weapon and the target, the posture sensor is used to collect spatial angle data of the weapon, and the camera is used to collect image data of the aiming direction; the image recognition module is used to obtain the image data collected by the camera and identify the human body and human body parts in the image data; the weapon fixture is used to fix the digital transmitter to weapons of different calibers; the weapon main control is used to obtain the position data collected by the RTK positioning module, the spatial angle data collected by the posture sensor, and the image data processed by the image recognition module, and form digital ammunition to be sent to the soldier terminal through the communication module; A soldier terminal is connected to the weapon main control and guidance control terminals of the digital transmitter, respectively, and is used to send digital ammunition to the guidance control terminal and obtain the injury determination result generated by the guidance control terminal; The guidance and control terminal performs trajectory deduction based on the position data and spatial angle data to obtain the basic trajectory, and then identifies the impact site according to the image data to correct the basic trajectory to obtain the final trajectory. Based on the final trajectory, the collision damage model is determined to generate an injury determination result.

9. The device according to claim 8, characterized in that The deviation of the position data obtained by the RTK positioning module through repeated positioning collection does not exceed 10 cm. The RTK positioning module adopts an integrated structure of the positioning module and the antenna, or adopts a split structure to integrate the RTK module into the weapon main control and lead out the antenna feeder; the attitude sensor includes a gyroscope, an electronic compass, an inclinometer and a dual-antenna directional module, and the collected spatial angle data includes pitch angle, roll angle and heading angle; the camera is selected from at least one of an ordinary camera, an infrared camera and a thermal imaging camera, and the focal length of the camera is adjustable, including optical zoom and digital zoom.

10. The device according to claim 8, characterized in that It also includes a flash simulator, a flame sensor and a split power supply structure. The flash simulator is used to simulate the flame when the weapon is fired, and the flame sensor is used to identify the flame generated during live ammunition shooting; the split power supply structure includes a host end and a battery that are separately set, and the battery is installed through a replaceable battery tube.

Images