Machine vision AI high-precision throwing target scoring system

Through machine vision AI technology, the problem of low recognition accuracy of throwing scores in military throwing training is solved, and a high-precision throwing target reporting system is realized, which improves training efficiency and provides accurate data analysis and action guidance.

CN118913026APending Publication Date: 2024-11-08SICHUAN MITRUI HUICHUANG TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410963085.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-18
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

The prior art is difficult to achieve high-precision recognition and assessment of throwing results in military throwing training, resulting in low training efficiency and large errors.

Method used

Using machine vision AI technology, the video stream and reference pictures of thrown objects are collected through the front-end data acquisition module. The back-end control module uses AI technology to calculate the number of rings and offsets of the thrown objects and the target to achieve high-precision throwing and target reporting. system.

Benefits of technology

It improves the accuracy and data-based assessment of throwing training, can analyze soldiers' throwing actions based on display data and provide commander comments. It is highly adaptable, fast layout and easy to use.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118913026A_ABST
    Figure CN118913026A_ABST
Patent Text Reader

Abstract

The invention relates to a machine vision AI high-precision throwing target scoring system, and the system comprises a front-end data collection module which is used for collecting a reference picture and a video stream of a process from a parabola of a throwing object to a bomb landing point of a throwing target, and discriminating a picture, falling on the throwing target, of the throwing object in the video stream based on an AI technology; and the rear-end control module is used for comparing the discriminated image with a reference picture, and calculating the ring number and the offset taking the center of the target as a reference to carry out throwing target scoring. The method can solve the problems of low soldier throwing training score identification precision, low assessment efficiency and the like at present.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of throwing training equipment for throwing weapons for military and police use, and in particular to a machine vision AI high-precision throwing target reporting system. Background Art

[0002] Throwing throwing weapons into the target range is an important military training project. During the throwing training process, according to the different distances of the throwing target, the terrain and objects are flexibly used, and the throwing of grenades and hand grenades is completed independently according to different methods such as bomb training subjects, and the size of its lethality is judged according to the distance between the throwing point and the target. The closer the distance between the throwing point and the target, the greater its lethality. At present, the assessment and judgment of domestic military throwing training results are mainly based on the manual determination of the distance between the throwing point and the target. Its assessment method makes it difficult to improve the training efficiency, and there are large errors, making it difficult to achieve accurate throwing training and assessment. Based on this, the present invention adopts a machine vision AI technology to realize a long-distance high-precision throwing target reporting system to solve the above problems and fill the gap in the field of high-precision throwing recognition systems. Summary of the invention

[0003] The purpose of the present invention is to provide a machine vision AI high-precision throwing target reporting system to solve the problems existing in the above-mentioned prior art, and to solve the current problems of low recognition accuracy and low assessment efficiency of soldiers' throwing training results.

[0004] To achieve the above object, the present invention provides the following solutions:

[0005] A machine vision AI high-precision throwing and reporting target system, comprising:

[0006] The front-end data acquisition module is used to collect the reference image and the video stream of the process of the parabola of the thrown object to the impact point of the thrown target, and identify the image of the thrown object landing on the thrown target in the video stream based on AI technology;

[0007] The back-end control module is used to compare the identified image with the reference image, calculate the number of rings and the offset based on the center of the target circle for throwing and reporting the target.

[0008] Optionally, the front-end data acquisition module includes:

[0009] A reference image acquisition submodule is used to acquire a throwing target image, correct the deformed changes produced by the throwing target image, determine the effective range of the throwing target, and generate a reference image;

[0010] The video stream acquisition submodule is used to collect the reference image and the video stream of the process from the parabola of the projectile to the impact point of the projectile target, and to identify the image of the contact point between the projectile point and the projectile target in the process of collecting the sound signal after the projectile explodes or lands in the video stream based on AI technology.

[0011] Optionally, the reference image acquisition submodule includes:

[0012] An area recognition unit is used to identify the corners of a throwing target, determine the position of the throwing target based on the corners, identify a preset ring number identification line in the throwing target based on the throwing target position, determine the center point of the throwing target, take the center point as the origin, take the origin as the base point, virtually set an X-axis and a Y-axis perpendicular to the edge of the throwing target, take the center point as the starting point, extend a number of virtual lines in the X-axis and Y-axis at preset angles, take the intersection of the X-axis, the Y-axis and the several virtual lines with the ring number identification line as the recognition point, and determine the effective recognition area in the throwing target through the recognition point.

[0013] Optionally, the reference image acquisition submodule further includes:

[0014] An effective range identification unit is used to collect the throwing target image, correct the deformed changes produced by the throwing target image based on a distortion correction algorithm, restore the image according to the identification points in the effective identification area, calculate the distance and position of the identification points in the throwing target image after image restoration, determine the effective range in the throwing target, and generate the reference image.

[0015] Optionally, the backend control module includes:

[0016] The back-end control submodule includes: using opencv projection transformation technology to restore the identified image, determine the bullet point or explosion trace, compare the bullet point or explosion trace with the reference image, use machine vision software technology to calculate the number of rings and offset based on the center of the target, transmit the number of rings and offset to the throwing terminal for graphical display, and complete the throwing target reporting.

[0017] Optionally, the back-end control module further includes: a reference image updating submodule, which is used to update the image obtained after the acoustic signal of the projectile explodes or falls to the ground to a new reference image.

[0018] Optionally, the back-end control module further includes: the back-end control submodule further includes: a bombing action acquisition unit, used for acquiring a bombing action image and transmitting the bombing action image to the throwing terminal.

[0019] The beneficial effects of the present invention are:

[0020] The present invention can solve the current problem of soldiers' throwing accuracy and data assessment;

[0021] The present invention can analyze the soldiers' throwing data according to the displayed data and provide it to the commander to comment on and correct the problems existing in each soldier's throwing action;

[0022] The invention can be used freely in a movable manner or in a fixed manner, and has the advantages of strong adaptability, quick deployment, and convenient use. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0024] Figure 1 A schematic diagram of a machine vision AI high-precision throwing and marking system according to an embodiment of the present invention;

[0025] Figure 2 A schematic diagram of a software algorithm according to an embodiment of the present invention;

[0026] Among them, 1-bombing target, 2-high-speed camera, 3-pan head, 4-receiving antenna, 5-bracket, 6-bracket base, 7-camera power supply, 8-throwing terminal tablet, 9-transmitting antenna, 10-throwing control host. DETAILED DESCRIPTION

[0027] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of 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 of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0028] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0029] like Figure 1 As shown, this embodiment discloses a machine vision AI high-precision throwing target reporting system, including: a throwing target 1, the reporting system also includes a front-end data acquisition module and a back-end control module, wherein the throwing target 1 is set to 25m according to the training subject 2square, and set the number of rings within this range. The diameter of the bull's eye ring is 1m, and the subsequent target ring lines increase by 0.5m in radius, namely: 10 rings, 9 rings, 8 rings, 7 rings, and 6 rings. A ring number identification line is formed between two adjacent target ring lines, and quadrant identification lines are formed in the X-axis direction and the Y-axis direction with the bull's eye as the center position; the front-end data acquisition module forms an inclination towards the throwing target, and is configured to automatically identify the target ring line, the ring number identification line and the quadrant identification line after receiving the throwing command, and use video stream to capture the process of the grenadier throwing and the parabola of the thrown object to the impact point of the throwing target, and collect the impact image according to the fixed focus command and the camera command, and use the opencv projection transformation technology to restore the image; the back-end control module and the front-end The data acquisition module is connected via wireless communication transmission and transmits the original image signal via wireless communication. The display control device includes an acquisition control module, an image processing module and a display module. Among them, the acquisition control module adopts video stream capture. When the projectile lands, the AI ​​technology recognizes the moment when the projectile lands, implements capture, and transmits the captured pictures to the throwing control host via wireless communication. The pictures are analyzed and processed by software in a high-performance computer, and the processed pictures are compared with the original pictures stored in the computer. Then, the deep learning soft method is used to realize the accurate calculation of the landing point of the projectile. After the calculation result, it is transmitted to the commander's throwing terminal tablet 8 or the throwing personnel's mobile phone via wireless signal to realize the target reporting function. The software analyzes images by first using a comparison algorithm. That is, before starting up and preparing to drop bombs, the acquisition end will take a blank photo as a reference photo. When the projectile falls on the target, the video stream captures an object falling to the ground. The software determines that an object has fallen to the ground and grabs this frame of photos to compare with the blank photo when the software was turned on to determine the valid photo. The deep learning algorithm is used to calculate the specific position of the landing point on the target. Taking the center point of the ten rings of the target as the reference, the number of rings and offset of the projectile are calculated to accurately calculate the number of rings thrown on the target.

[0030] The front-end data acquisition processing module may include an image processing submodule, an image formation submodule and a target distortion correction submodule, wherein the image processing submodule may be configured to acquire the impact target, and send the impact image to the image formation submodule, and send the image acquired after power-on to the target distortion correction submodule, so as to realize automatic tracking and locking of the throwing target; the image formation submodule may be configured to determine the impact position based on the impact image, calculate the number of impact point rings and the center of the target offset of the current throw; the target distortion correction submodule may be configured to automatically retrieve the target ring line, the ring number identification line and the quadrant identification line through the opencv projection transformation technology, and output the center of the target correction information.

[0031] The display control device may also include a display module, which is connected to the image forming submodule via wireless signal transmission and is configured to graphically display the number of impact point rings and the center of the target offset for each throw.

[0032] The front-end data acquisition mechanism may include a high-speed camera, a receiving and transmitting antenna 4, a pan-tilt head 3, a bracket 5, a bracket base 6, a camera and a camera power supply 7; the high-speed camera is set on the pan-tilt head 3; the pan-tilt head 3 is set on the bracket 5; the bracket 5 is set on the bracket base 6; the receiving and transmitting antenna 9 is connected to the high-speed camera 2; the camera and the camera power supply 7 are connected to the high-speed camera 2 and the pan-tilt head 3, and are configured to provide power to the high-speed camera 2 and the pan-tilt head 3.

[0033] The front-end data acquisition mechanism may also include a first signal transmission device connected to the power supply control device, and the back-end control structure may also include a second signal transmission device connected to the display control device. The first signal transmission device and the second signal transmission device are connected to form a dual-loop signal connection to realize the transmission of throwing instructions, camera instructions, focus instructions and picture acquisition instructions.

[0034] This embodiment also discloses a workflow of a machine vision AI high-precision throwing and marking system, including:

[0035] like Figure 2 As shown, when in use, after the high-speed camera 2 is turned on, it actively takes a photo and transmits it to the throwing control host 10. The deformed changes caused by the camera taking the picture are corrected through machine vision AI technology, and the effective range of the target paper is determined as the recognition reference, and the recognition reference image is stored in the throwing control host 10 to realize the reference of the deep learning algorithm of the comparison algorithm software, and complete the startup preparation work before throwing, which is specifically:

[0036] Step 1: Region Identification:

[0037] Identify the four corners of the target and determine the target position. Then identify the three circles of 10 rings, 9 rings and 8 rings to determine the center point of the target. Then use the center point of the target as the origin and virtually set up the X-axis and Y-axis perpendicular to the edge of the target. The X-axis and Y-axis virtually divide the target into four quadrants. Then use the center point of the target as the starting point and extend four virtual lines at 45° on the X-axis or Y-axis. Divide the target into eight parts.

[0038] The virtual Y-axis and four virtual lines extending at 45° to the X-axis or Y-axis, the intersection points with the target ring number and the intersection points with the edge of the target are used as identification points.

[0039] The target position and the effective recognition area are determined by eight recognition points distributed on the edge of the target and eight recognition points distributed on the ten-ring line.

[0040] Step 2: Detection method:

[0041] By identifying a total of 32 unique identification points distributed on the ten-ring, nine-ring, eight-ring and target edge, the target is imaged and identified at the bottom layer, and the distortion correction algorithm corrects the image distortion caused by the angle between the target and the camera, thereby achieving image restoration.

[0042] By dividing the quadrants and using 24 identification points in addition to the eight points on the edge of the target, the algorithm is used to calculate the distance and position. When the impact point is within the effective range, accurate target reporting can be achieved through deep learning algorithm software with an accuracy of up to 0.5 mm.

[0043] Further elaboration:

[0044] Step 1: Check the ten rings, nine rings, and eight rings to determine the center of the target paper.

[0045] Step 2: Detect the eight symmetrical points 1.1, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7, and 1.8 to determine the effective range of the target's impact point.

[0046] Step 3: Detect the 24 points where the ring numbers 2.1-5.6 intersect with the quadrant area dividing lines, and determine the physical distance and position of the 24 intersection points between the quadrant dividing lines and the ring lines.

[0047] When the projectile impact point is within the effective range, the deep learning algorithm software can be used to accurately report the target. Through experiments, the accuracy of the present invention can reach 0.5 mm.

[0048] The throwing terminal tablet 8 sends a bombing command to the high-speed camera 2. The high-speed camera 2 uses video stream to capture the process of the grenadier throwing and the parabola of the thrown object to the impact point of the throwing target. The video stream uses AI technology to identify the picture of the thrown object falling on the throwing target. The video stream formed in the process of collecting sound signals after the thrown object explodes or falls to the ground identifies multiple pictures of the contact point between the bomb point and the throwing target. The picture identified by the high-speed camera 2 is transmitted to the throwing control host 10 through the receiving antenna 4 to complete the front-end data collection.

[0049] The data processing mechanism in the throwing control host 10 uses the opencv projection transformation technology to restore the image, and compares it with the reference image stored in the throwing control host 10 according to the impact point or explosion trace. These features determine the limited range and number of rings of the target, and use machine vision software technology to calculate the precise number of rings and offset based on the center of the target circle, and at the same time update the camera image triggered by the sound signal to a new reference image. If the parabola impact point image collected by the high-speed camera 2 is outside the 1-throwing target, the throwing control host 10 is judged as off-target; if the thrown object explodes before it lands, the throwing control host 10 is judged as air explosion, and the precise number of rings and offset based on the center of the target circle calculated by the machine vision software technology and the student's throwing action image are transmitted to the distant throwing terminal tablet 8 through the communication system to realize the throwing target reporting function, and at the same time provide more scientific throwing action guidance through AI body capture.

[0050] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A machine vision AI high-precision throwing and reporting target system, characterized in that: include: The front-end data acquisition module is used to collect the reference image and the video stream of the process of the parabola of the thrown object to the impact point of the thrown target, and identify the image of the thrown object landing on the thrown target in the video stream based on AI technology; The back-end control module is used to compare the identified image with the reference image, calculate the number of rings and the offset based on the center of the target circle for throwing and reporting the target.

2. The machine vision AI high-precision throwing target reporting system according to claim 1 is characterized in that: The front-end data acquisition module includes: A reference image acquisition submodule is used to acquire a throwing target image, correct the deformed changes produced by the throwing target image, determine the effective range of the throwing target, and generate a reference image; The video stream acquisition submodule is used to collect the reference image and the video stream of the process from the parabola of the projectile to the impact point of the projectile target, and to identify the image of the contact point between the projectile point and the projectile target in the process of collecting the sound signal after the projectile explodes or lands in the video stream based on AI technology.

3. The machine vision AI high-precision throwing target reporting system according to claim 2 is characterized in that: The reference image acquisition submodule includes: An area recognition unit is used to identify the corners of a throwing target, determine the position of the throwing target based on the corners, identify a preset ring number identification line in the throwing target based on the throwing target position, determine the center point of the throwing target, take the center point as the origin, take the origin as the base point, virtually set an X-axis and a Y-axis perpendicular to the edge of the throwing target, take the center point as the starting point, extend a number of virtual lines in the X-axis and the Y-axis at preset angles, take the intersection of the X-axis, the Y-axis and the several virtual lines with the ring number identification line as the recognition point, and determine the effective recognition area in the throwing target through the recognition point.

4. The machine vision AI high-precision throwing target reporting system according to claim 3 is characterized in that: The reference image acquisition submodule also includes: An effective range identification unit is used to collect the throwing target image, correct the deformed changes produced by the throwing target image based on a distortion correction algorithm, restore the image according to the identification points in the effective identification area, calculate the distance and position of the identification points in the throwing target image after image restoration, determine the effective range in the throwing target, and generate the reference image.

5. The machine vision AI high-precision throwing target reporting system according to claim 2 is characterized in that: The back-end control module includes: The back-end control submodule includes: using opencv projection transformation technology to restore the identified image, determine the bullet point or explosion trace, compare the bullet point or explosion trace with the reference image, use machine vision software technology to calculate the number of rings and offset based on the center of the target, transmit the number of rings and offset to the throwing terminal for graphical display, and complete the throwing target reporting.

6. The machine vision AI high-precision throwing target reporting system according to claim 5, characterized in that: The back-end control module further includes: a reference image updating submodule, which is used to update the image obtained after the acoustic signal collection is completed after the projectile explodes or the projectile falls to the ground to a new reference image.

7. The machine vision AI high-precision throwing target reporting system according to claim 5, characterized in that: The back-end control module also includes: The back-end control submodule also includes: a bombing action acquisition unit, which is used to acquire a bombing action image and transmit the bombing action image to the throwing terminal.