CR-39 detector alpha trajectory counting system and method based on YOLOv8 algorithm
Through the CR-39 detector alpha trajectory counting system of YOLOv8 algorithm, an automated alpha trajectory counting is realized, solving the time-consuming and cost-effective problems in the existing technology, and providing efficient and accurate counting results and simplified operational processes.
Patent Information
- Application Number
- CN202510259543.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-06
- Publication Date
- 2025-07-01
AI Technical Summary
The existing CR-39 detector alpha trajectory counting method relies on manual counting, is time-consuming and labor-intensive, susceptible to visual fatigue, and is expensive to use in instruments and equipment.
The CR-39 detector alpha trajectory counting system based on the YOLOv8 algorithm is used, combining a mobile platform, an electronic eyepiece, a microscope, a computer and a microcontroller, and uses a deep convolutional neural network for automated image analysis and counting.
Fast and accurate alpha trajectory counting is achieved, reducing equipment costs, simplifying operational processes, and supporting data tracking and analysis.
Smart Images
Figure CN120233392A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to nuclear radiation detection technology, in particular to a CR-39 detector alpha track counting system and method based on the YOLOv8 algorithm. Background Art
[0002] The CR-39 detector has high sensitivity to radon and other radioactive elements. When exposed to an environment rich in radon, the radioactive particles generated by the decay of radon will leave tracks on the CR-39 detector. The tracks can be magnified and observed under an optical microscope. By calculating the number of alpha tracks on the CR-39 detector in combination with the actual area and measurement time of the detector, the concentration of radon can be obtained.
[0003] The number of tracks on the CR-39 detector changes with the concentration and exposure time. The existing traditional counting method relies on manual counting under a microscope, which is time-consuming and laborious. Especially when the number of tracks is large and the density is high, it is easy to cause visual fatigue; and manually adjusting the microscope may cause position deviation and vibration, significantly affecting the counting result and ultimately affecting the evaluation of radon concentration. For this reason, automatic track measurement methods for CR-39 using Hamamatsu C-1285 multi-processor image analysis systems, automatic sliding scanning systems based on track shape analysis, photometric methods for measuring the track density of SSNTDs, and rapid measurement methods such as measuring alpha exposure on CR-39 detectors using ultraviolet-visible spectrophotometry have been developed. Currently, there are many advanced products on the market, such as Autoscan60 developed by Thermo Fisher Scientific (Santa Fe, MN, USA), the Radometer 2000 series launched by Radosys Ltd. (Budapest, Hungary), the Taslimage system developed by Track Analysis Systems Ltd. (Bristol, U.K.), and HSP-1000 produced by Seiko Precision Inc. (Chiba, Japan). However, the instruments involved in these methods and the products on the market are still relatively expensive. The YOLOv8 algorithm uses a deep convolutional neural network to divide an image into grids, and each grid predicts bounding boxes and class probabilities, thus achieving fast and accurate object detection. Currently, the YOLOv8 algorithm has been widely applied in multiple fields, such as the identification and monitoring of leaf diseases, object detection in drone aerial images, and real-time object detection in side-scan sonar images, etc.; the recognition of alpha track images can be further improved through the YOLOv8 algorithm. Therefore, it is possible to combine the YOLOv8 algorithm with the CR-39 detector to improve alpha track counting during the radon concentration measurement. Summary of the Invention
[0004] The object of the present invention is to overcome the above deficiencies of the prior art and provide a CR-39 detector alpha track counting system and method based on the YOLOv8 algorithm.
[0005] The technical solution of the present invention is: A CR-39 detector alpha track counting system based on the YOLOv8 algorithm, including a mobile platform, an electronic eyepiece, a microscope, a host computer and a microcontroller.
[0006] The mobile platform includes a left-right movement module, a front-back movement module, a stage and a communication module. The left-right movement module includes a base, a moving lead screw, a slider and a stepping motor. The slider is sleeved on the moving lead screw, and the moving lead screw and the slider are arranged on the base. The output shaft of the stepping motor is connected to the moving lead screw. Driven by the stepping motor, the slider slides on the base through the rotation of the moving lead screw. The structure and movement mode of the front-back movement module are the same as those of the left-right movement module; The base of the front-back movement module is fixedly installed on the slider of the left-right movement module, and the stage is fixedly installed on the slider of the front-back movement module.
[0007] The electronic eyepiece is fixedly installed at the top of the microscope. The microscope is placed on one side of the mobile platform, and the lens of the microscope is located above the stage of the mobile platform.
[0008] The host computer includes an image acquisition module, an image storage module, a YOLOv8 algorithm module, a mobile platform path preset module and a communication module. The image acquisition module, the image storage module and the YOLOv8 algorithm module are electrically connected in sequence; the image acquisition module is used to capture images of alpha tracks on the CR-39 detector to be measured, the image storage module is used to store the alpha tracks on the captured CR-39 detector, and to splice all the captured images for easy verification. The YOLOv8 algorithm module is used to quickly analyze and identify and count various alpha track images on the captured CR-39 detector. The mobile platform path preset module is set with the moving path of the mobile platform to control the front-back, left-right movement of the mobile platform; the microcontroller is respectively connected to the communication module of the mobile platform and the communication module of the host computer.
[0009] During use, the CR-39 detector to be counted is placed on the stage of the mobile platform. Under the action of the microcontroller, the stepping motor of the mobile platform drives the stage to move back and forth, left and right. At the same time, the host computer captures, analyzes and counts the alpha tracks of the CR-39 detector recorded by the electronic eyepiece.
[0010] A further technical solution of the present invention is that the YOLOv8 algorithm module is trained with more than 80,000 diverse alpha track samples, and a convolutional neural network is used to quickly analyze images; the microcontroller sends the moving path of the mobile platform preset in the upper computer mobile platform path preset module to the microcontroller, and the microcontroller controls the stepping motors in the left-right moving module and the front-back moving module of the mobile platform to act, thereby driving the loading platform to move forward, backward, left and right.
[0011] A further technical solution of the present invention is that the moving path of the moving platform 1 adopts an automatic mode and moves in an "S" shape.
[0012] Another technical solution provided by the present invention is a method applied to the aforementioned CR-39 detector alpha track counting system based on the YOLOv8 algorithm, including the following steps: I. Image acquisition Start the mobile platform, place the CR-39 detector to be counted on the loading platform of the mobile platform, and the upper computer communication module sends the moving path of the mobile platform preset in the mobile platform path preset module to the microcontroller. The microcontroller controls the movement of the stepping motor of the mobile platform electrically connected to it, and the stepping motor drives the loading platform on the mobile platform to move in an "S" shape; at the same time, the image acquisition module of the upper computer synchronously acquires the images of the CR-39 detector imaging and stores them in the image storage module in sequence.
[0013] II. Automatic recognition and counting Put the images in the image storage module into the YOLOv8 algorithm module trained in advance with a large number of track samples. In this model, a deep convolutional neural network is used to divide the image into grids, and each grid predicts the bounding box and class probability to achieve fast and accurate track target detection. The images after detection are saved to a new path in the image storage module 4-2, and the tracks in the images are marked on the images. After all the images are recognized, the YOLOv8 algorithm module counts the total count, and at the same time splices all the images in the acquisition order and restores them to the CR-39 detector image.
[0014] A further technical solution of the present invention is that before the CR-39 detector to be counted is placed on the loading platform, it is pre-treated, specifically by placing it in a 6.25M potassium hydroxide solution and etching it at 80°C for 10-12 hours.
[0015] The present invention has the following characteristics compared with the prior art: 1. The system of the present invention is more cost-effective, making the product price more affordable.
[0016] 2. The user interface of the system of the present invention is intuitive, the operation process is simplified, and it is easy to get started.
[0017] 3. The system of the present invention adopts the advanced YOLOv8 algorithm to ensure high accuracy of the results.
[0018] 4. The method of the present invention can achieve data tracking and analysis, supporting subsequent research.
[0019] The detailed structure of the present invention will be further described below in conjunction with the accompanying drawings and specific embodiments. Description of the Drawings
[0020] Figure 1 It is a schematic structural diagram of the alpha track counting system of the CR-39 detector; Figure 2 It is a schematic structural connection diagram between various parts of the mobile platform; Figure 3 It is a schematic connection diagram between various modules inside the host computer; Figure 4 It is a linear fitting comparison diagram of the model count value Vm and the corrected count value Vc. Specific Embodiments
[0021] Example 1, as Figures 1 - 4 shown, a CR-39 detector alpha track counting system based on the YOLOv8 algorithm includes a mobile platform 1, an electronic eyepiece 2, a microscope 3, a host computer 4, and a microcontroller 5.
[0022] The mobile platform 1 includes a left-right movement module 1-1, a front-back movement module 1-2, a stage 1-3, and a communication module (not shown in the figure). The left-right movement module 1-1 includes a base 1-1-1, a moving lead screw 1-1-2, a slider 1-1-3, and a stepper motor 1-1-4. The slider 1-1-3 is sleeved on the moving lead screw 1-1-2. The moving lead screw 1-1-2 and the slider 1-1-3 are arranged on the base 1-1-1. The output shaft of the stepper motor 1-1-4 is connected to the moving lead screw 1-1-2. Driven by the stepper motor 1-1-4, the rotation of the moving lead screw 1-1-2 drives the slider 1-1-3 to slide on the base 1-1-1, realizing the left-right movement of the slider 1-1-3. The structure and movement mode of the front-back movement module 1-2 are the same as those of the left-right movement module 1-1. The base of the front-back movement module 1-2 is fixedly installed on the slider 1-1-3 of the left-right movement module 1-1, and the stage 1-3 is fixedly installed on the slider of the front-back movement module 1-2, so that the stage 1-3 can move arbitrarily in the front-back and left-right directions.
[0023] The electronic eyepiece 2 is fixedly installed at the top of the microscope 3. The microscope 3 is placed on one side of the mobile platform 1, and the lens of the microscope 3 is located above the stage 1-3 of the mobile platform 1.
[0024] The host computer 4 includes an image acquisition module 4-1, an image storage module 4-2, a YOLOv8 algorithm module 4-3, a mobile platform path preset module 4-4, and a communication module (not shown in the figure). The image acquisition module 4-1, the image storage module 4-2, and the YOLOv8 algorithm module 4-3 are electrically connected in sequence. The image acquisition module 4-1 is used to capture images of alpha tracks on the CR-39 detector to be measured. For a CR-39 detector sample with a standard size of usually 1 cm × 1 cm, a total of 910 images will be generated when using the image acquisition module 4-1 to capture tracks. The image storage module 4-2 is used to store the alpha tracks on the captured CR-39 detector and splice all the captured images for easy verification. The YOLOv8 algorithm module 4-3 is used to quickly analyze and identify and count various alpha track images on the captured CR-39 detector. It is trained with more than 80,000 diverse alpha track samples and uses a convolutional neural network to quickly analyze the images. The mobile platform path preset module 4-4 is set with a mobile platform movement path for controlling the front, back, left, and right movement of the mobile platform 1.
[0025] Due to the influence of the movement accuracy of the mobile platform 1, if the mobile platform 1 first moves forward and backward and then moves left and right, or vice versa, there may be a large error in the captured images. Therefore, the movement path of the mobile platform 1 is set to an automatic mode and moves in an "S" shape to reduce the movement error during the image capture process. The inherent accuracy error of the stepper motor and the lead screw is relatively small and can be ignored during the counting process.
[0026] The microcontroller 5 is respectively connected to the communication module of the mobile platform 1 and the communication module of the host computer 4, and sends the mobile platform movement path preset in the mobile platform path preset module 4-4 of the host computer 4 to the microcontroller 5. The microcontroller 5 controls the stepper motors in the left and right movement module 1-1 and the front and back movement module 1-2 of the mobile platform, thereby driving the carrier table 1-3 to move forward, backward, left, and right.
[0027] When in use, the CR-39 detector to be counted is placed on the carrier table 1-3 of the mobile platform 1. Under the action of the microcontroller 5, the stepper motor of the mobile platform 1 drives the carrier table 1-3 to move forward, backward, left, and right. At the same time, the host computer 4 captures, analyzes, and counts the alpha tracks of the images recorded by the electronic eyepiece 2 of the CR-39 detector.
[0028] Embodiment 2. A method for a CR-39 detector alpha track counting system based on the YOLOv8 algorithm described in Embodiment 1 includes the following steps: 1. Image acquisition Start the mobile platform 1, place the CR-39 detector to be counted on the stage 1-3 of the mobile platform 1. The communication module of the host computer 4 sends the preset moving path of the mobile platform in the mobile platform path preset module 4-4 to the microcontroller 5. The microcontroller 5 controls the movement of the stepper motor of the mobile platform 1 electrically connected to it, and the stepper motor drives the stage 1-3 on the mobile platform 1 to move in an S shape. At the same time, the image acquisition module 4-1 of the host computer 4 displays the imaging of the CR-39 detector of the microscope 3 and the electronic eyepiece 2 in real time, synchronously acquires the images of the CR-39 detector imaging, and stores the 910 images captured by the CR-39 detector into the image storage module 4-2 in sequence. In this embodiment, 16 CR-39 detectors to be counted are set. These detectors have been used in experiments to measure granite, cement floors, and radon sources before. The measurement time and radon concentration values of these CR-39 detectors vary due to the specific settings of their respective experiments, and the alpha tracks of the 16 CR-39 detectors to be counted are different.
[0029] To further improve the accuracy of alpha track counting, before placing the CR-39 detector to be counted on the stage 1-3, preprocess the CR-39 detector to be counted. Specifically, place it in a 6.25M potassium hydroxide solution and etch it at 80°C for 10 to 12 hours. The longer the etching time, the more obvious the tracks. After etching, clean and wipe it, which can facilitate the recognition of the YOLOv8 algorithm module 4-3, because larger and cleaner targets can usually enable YOLOv8 to capture features more effectively and improve the recognition and positioning effects.
[0030] II. Automatic recognition and counting Put the images in the image storage module 4-2 into the YOLOv8 algorithm module 4-3 that has been trained with a large number of track samples in advance. In this model, use a deep convolutional neural network to divide the image into grids, and each grid predicts the bounding box and class probability to achieve fast and accurate track target detection. The images after detection are saved to a new path in the image storage module 4-2, and the tracks in the images are marked on the images. After all the images are recognized, the YOLOv8 algorithm module 4-3 counts the total count and records it as the model count value Vm. At the same time, splice these 910 images in the acquisition order to restore the CR-39 detector image. Identify and count the alpha tracks on 16 CR-39 detectors to be counted respectively.
[0031] To verify the accuracy of the total count model count value Vm in the YOLOv8 algorithm module 4-3, 16 restored CR-39 detector images were manually inspected to check for omissions or misjudgments in the images, and the model count value Vm was corrected. The corrected count value was denoted as Vc. The percentage error between the model count value Vm and the corrected count value Vc was calculated as the evaluation index for the total count statistics of the YOLOv8 algorithm module 4-3, and the results shown in Table 1 below were obtained.
[0032] Table 1: Model count values, corrected count values, and percentage errors of CR-39 detectors CR-39 Detector Number <![CDATA[Model count value (V m )]]> <![CDATA[Corrected count value (V c )]]> Percentage Error CR-39 Detector Number <![CDATA[Model count value (V m )]]> <![CDATA[Corrected count value (V c )]]> Percentage Error NO.1 1966 1978 0.61% NO.9 9606 9683 0.77% NO.2 3230 3236 0.18% NO.10 11239 11233 -0.05% NO.3 3576 3559 -0.48% NO.11 12503 12656 1.22% NO.4 3747 3748 0.02% NO.12 13203 13219 0.12% NO.5 4607 4695 1.91% NO.13 13273 13456 1.38% NO.6 4787 4819 0.67% NO.14 16448 16798 2.13% NO.7 6555 6560 0.08% NO.15 17452 17858 2.33% NO.8 6946 7040 1.35% NO.16 20287 20280 -0.03% It can be seen from Table 1 that there are differences between the model count values and the calibrated count values of the selected 16 CR-39 detectors at different confidence levels. In this study, the model was trained with 80,000 samples, and the model count values were relatively close to the calibrated count values, with their errors generally remaining below 3%. Additionally, the change in the number of tracks from 1966 to 20287 had little impact on the accuracy of the total count statistics of the YOLOv8 algorithm module 4-3. It was highly robust to changes in track density and could adapt to changes within a certain range.
[0033] To visually analyze and apply these data, the Origin software was used to perform linear fitting on the model count value Vm and the corrected count value Vc respectively, as Figure 4 shown. It can be seen from Figure 4 that the R 2 value of the fitting curve of the model count value Vs and the corrected count value Vc was 0.99964. Thus, the fitting results clearly demonstrated the linear relationship between the data, indicating a strong correlation between the model count value Vm and the corrected count value Vc.
Claims
1. The CR-39 detector alpha track counting system based on the YOLOv8 algorithm is characterized by: It includes a mobile platform, an electronic eyepiece, a microscope, a host computer and a microcontroller; The mobile platform includes a left-right moving module, a front-back moving module, a loading platform and a communication module. The left-right moving module includes a base, a moving screw, a slider and a stepper motor. The slider is sleeved on the moving screw. The moving screw and the slider are arranged on the base. The output shaft of the stepper motor is connected to the moving screw. Under the drive of the stepper motor, the slider is driven to slide on the base through the rotation of the moving screw. The structure and movement mode of the front-back moving module are the same as those of the left-right moving module. The base of the forward and backward moving module is fixedly installed on the slider of the left and right moving module, and the loading platform is fixedly installed on the slider of the forward and backward moving module; The electronic eyepiece is fixedly mounted on the top of the microscope, the microscope is placed on one side of the mobile platform, and the lens of the microscope is located on the stage of the mobile platform; The host computer includes an image acquisition module, an image storage module, a YOLOv8 algorithm module, a mobile platform path preset module and a communication module, and the image acquisition module, the image storage module and the YOLOv8 algorithm module are electrically connected in sequence; the image acquisition module is used to capture images of the alpha track on the CR-39 detector to be tested, the image storage module is used to store the captured alpha track on the CR-39 detector, and to splice all captured images for easy verification, the YOLOv8 algorithm module is used to quickly analyze the captured CR-39 detector Various types of alpha track images are identified and counted, and the mobile platform path preset module is provided with a mobile platform moving path for controlling the forward, backward, left and right movement of the mobile platform; the microcontroller is respectively connected to the communication module of the mobile platform and the communication module of the host computer; When in use, the CR-39 detector to be counted is placed on the stage of the mobile platform. Under the action of the microcontroller, the stepper motor of the mobile platform drives the stage to move forward, backward, left and right. At the same time, the host computer captures, analyzes and counts the alpha track of the CR-39 detector in the image recorded by the electronic eyepiece.
2. The CR-39 detector alpha track counting system based on the YOLOv8 algorithm as claimed in claim 1, characterized in that: The YOLOv8 algorithm module is trained with more than 80,000 diverse alpha track samples and uses a convolutional neural network to quickly analyze images; the microcontroller sends the mobile platform movement path preset in the host computer mobile platform path preset module to the microcontroller, and the microcontroller controls the stepper motor movements in the mobile platform left and right movement module and the front and back movement module, thereby driving the stage to move forward, backward, left and right.
3. The CR-39 detector alpha track counting system based on the YOLOv8 algorithm as claimed in claim 1, characterized in that: The moving path of the mobile platform 1 adopts automatic mode and moves in an "S" shape.
4. The method applied to the CR-39 detector alpha track counting system based on the YOLOv8 algorithm as described in claims 1-3, characterized in that: The following steps are included:
1. Image acquisition The mobile platform is started, and the CR-39 detector to be counted is placed on the stage of the mobile platform. The upper computer communication module sends the mobile platform moving path preset in the mobile platform path preset module to the microcontroller. The microcontroller controls the movement of the mobile platform stepper motor electrically connected to it, and the stepper motor drives the stage on the mobile platform to move in an S shape. At the same time, the image acquisition module of the upper computer synchronously acquires the image of the CR-39 detector and stores it in the image storage module in sequence.
2. Automatic recognition and counting The images in the image storage module are placed in the YOLOv8 algorithm module that has been trained in advance using a large number of track samples. In this model, a deep convolutional neural network is used to divide the image into grids, and each grid predicts the bounding box and category probability to achieve fast and accurate track target detection. The detected images are saved to a new path in the image storage module 4-2, and the tracks in the image are marked on the image. After all images are recognized, the YOLOv8 algorithm module counts the total count and stitches all images in the order of acquisition to restore them to the CR-39 detector image.
5. The method for applying to the CR-39 detector alpha track counting system based on the YOLOv8 algorithm as claimed in claim 4, characterized in that: Before the CR-39 detector to be counted is placed on the stage, the CR-39 detector to be counted is pretreated, specifically, it is placed in a 6.25M potassium hydroxide solution and etched at 80°C for 10 to 12 hours.