An artificial intelligence identification and analysis method and system for airborne pollen

Abstract

The application relates to an artificial intelligence identification and analysis method and system for air-borne pollen, and relates to the technical field of pollen identification. The system comprises a PC, a rack, a camera one, a camera two, a lifting seat, an agar block containing pollen, a placing table for placing the agar block, and a sliding plate which is horizontally and slidably connected to the lifting seat. The lower end of the sliding plate is fixed with a cutter which is horizontally arranged. The camera one is located outside the side wall of the agar block and horizontally faces the agar block, and the camera two is located above the agar block. In the application, agar is used as the carrier of pollen. The pollen is not prone to stacking on the same plane, and the pollen is in a three-dimensional suspended state in the agar block, so that the flattening and deformation of the pollen are avoided, and the pollen is convenient to identify. In the image obtained by shooting each time, the pollen in the lower layer of the agar block is not included, and the pollen in the image obtained by shooting is not prone to stacking, so that the stacking of the pollen is greatly reduced, and the identification rate is improved. Through an AI data analysis model, the identification speed and the identification rate are improved.

Description

Technical Field

[0001] This application relates to the field of pollen identification technology, and in particular to an artificial intelligence identification and analysis method and system for airborne pollen. Background Technology

[0002] Pollen identification is a method for studying pollen morphology and used to identify plant species and groups. The main pollen group is airborne pollen. This method primarily involves observing the morphological characteristics of pollen, detecting, analyzing, and classifying pollen samples. Currently, image recognition technology is widely used for pollen identification, and with the rise of artificial intelligence, AI is also beginning to be applied in the field of pollen identification.

[0003] Patent application CN118794848A discloses an artificial intelligence-based laser scattering pollen grain size distribution detection method and system, comprising the following steps: creating an initial training dataset and processing the initial training dataset to obtain a model training dataset; creating an initial intelligent model and training the initial intelligent model using the model training dataset to obtain an intelligent detection model; irradiating the pollen to be tested with a laser generator and collecting the scattered light signal to obtain the light intensity information of the pollen to be tested; analyzing and processing the light intensity information of the pollen to be tested using the intelligent detection model to obtain grain size distribution data; and obtaining the pollen grain size distribution based on the grain size distribution data. This invention addresses the problem that existing pollen grain size distribution detection technologies cannot accurately and fully detect pollen grain size distribution characteristics while ensuring the accuracy of pollen grain size detection by combining artificial intelligence technology with laser light scattering.

[0004] Regarding the aforementioned technologies, the general operating method is to place a glass slide containing pollen into an analyzer for analysis. However, the pollen is unevenly distributed on the glass slide, which can easily cause it to overlap. The pollen is also easily deformed by compression, resulting in low recognition rate or inability to recognize some pollen. Summary of the Invention

[0005] This application provides an artificial intelligence identification and analysis method and system for airborne pollen. By using agar as a pollen carrier, the pollen layering and deformation are significantly reduced, and the recognition rate is improved by combining artificial intelligence technology.

[0006] This application provides an artificial intelligence-based identification and analysis system for airborne pollen, which adopts the following technical solution:

[0007] An artificial intelligence identification and analysis system for airborne pollen includes a PC, a frame, a first camera, a second camera, a lifting base, an agar block containing pollen, and a placement platform for placing the agar block. A sliding plate is horizontally connected to the lifting base. The lifting base is driven to rise and fall by a linear module, and the sliding plate is driven to slide by a linear module. A cutter is fixed to the lower end of the sliding plate and is horizontally positioned. The first camera is located outside the side wall of the agar block and horizontally facing the agar block. The second camera is located above the agar block and vertically facing the agar block. The signal output terminals of the first and second cameras are connected to the PC, and the PC controls the operation of the linear modules through a controller.

[0008] By employing the above technical solution, using agar as a pollen carrier, camera one captures an image of the pollen distribution within the agar layer. This image, processed by image recognition software on a PC, determines the positions of multiple horizontal cuts, ensuring these cuts avoid the pollen. Then, based on the height of the horizontal cuts, the height of the lifting seat is adjusted by controlling the operation of linear module one, thus adjusting the cutter to the corresponding cutting position. The cutter then moves horizontally to cut the upper layer of the agar block. After cutting, the process pauses; at this point, the cut agar layer lies on the cutter, and the cutter's separation prevents camera two from capturing the lower pollen layers. The image captured by camera two is input into the PC, where image recognition software identifies, classifies, and statistically analyzes the pollen.

[0009] Optionally, the frame is fixed with a side light, which is located on the side of the agar block away from the camera and is horizontally oriented towards the agar block.

[0010] By adopting the above technical solution, with the side lamp and the camera positioned on the backlight side of the agar block, the camera can clearly capture an image of the pollen distribution within the agar block.

[0011] Optionally, the lifting seat is fixed with a baffle, which is vertically arranged to contact the side wall of the agar block, and the top of the baffle corresponds to the bottom of the cutter at the same height.

[0012] By adopting the above technical solution, when the cutter cuts the agar block, the baffle plate supports the right side of the agar block, ensuring that the cutter can cut the upper layer of the agar block while the lower layer of the agar block does not move. The height of the baffle plate rises and falls with the lifting seat, which can achieve the best blocking effect on the lower layer of the agar block.

[0013] Optionally, the placement platform is slidably connected to the frame, and the placement platform is driven to slide by a linear module.

[0014] By adopting the above technical solution, the electric moving placement platform moves to the outside and the top is unobstructed, which facilitates the placement of agar blocks.

[0015] Optionally, the frame is equipped with a three-dimensional moving platform, and the second camera is fixed to the movable end of the three-dimensional moving platform.

[0016] By adopting the above technical solution, the position of the camera in space can be precisely adjusted through a three-dimensional moving platform, thereby increasing the freedom of shooting position.

[0017] Optionally, a conveyor belt mechanism is also included, located on the side of the agar block away from the slide plate, and the conveyor belt mechanism is used to place the cut agar layer.

[0018] By adopting the above technical solution, the conveyor belt mechanism is used to place the cut agar layers, and each agar layer is conveyed backward a certain distance. The agar layers are arranged on the conveyor belt mechanism, which facilitates subsequent management and meets the need for reuse of the agar layers.

[0019] Optionally, the lifting seat is equipped with an electric cylinder and a second baffle that is driven to rise and fall by the electric cylinder. When the slide plate slides to the right limit, the second baffle is used to insert between the slide plate and the agar layer.

[0020] By adopting the above technical solution, when the slide plate slides to the right limit, the agar layer on the cutter is located directly above the conveyor belt mechanism. Then, the electric cylinder drives the second baffle to move down. The second baffle is used to insert between the slide plate and the agar layer. Then, the slide plate moves to the left and resets. Through the obstruction of the second baffle, the agar layer loses the support of the cutter and falls down onto the conveyor belt mechanism.

[0021] Secondly, this application provides an artificial intelligence-based method for identifying and analyzing airborne pollen, employing the following technical solution:

[0022] An artificial intelligence-based method for identifying and analyzing airborne pollen, using the aforementioned artificial intelligence-based system for identifying and analyzing airborne pollen, includes the following steps:

[0023] Step S1: Make agar blocks containing pollen using a mold;

[0024] Step S2: Place the agar block on the placement platform, start camera one, and the camera one takes an image of the pollen distribution inside the agar block. This image is used by image recognition software in the PC to determine the positions of multiple horizontal cuts, with the cut positions avoiding the pollen.

[0025] Step S3: Based on the height position of the horizontal cut, adjust the height position of the lifting seat by controlling the operation of the linear module one until the cutter reaches the corresponding cutting position. Then the linear module two works, and the cutter moves horizontally to cut the upper layer of the agar block into an agar layer. After the cutting is completed, pause and the lifting seat moves up to bring the agar layer closer to the camera two.

[0026] Step S4: The second camera captures an image of the agar layer and sends it to the PC. Using image recognition software combined with an AI data analysis model, the pollen is identified, classified, and counted.

[0027] Step S5: Remove the agar layer from the cutter, reset the cutter, and repeat steps S3-S4.

[0028] Optionally, the AI ​​data analysis model includes a model training dataset. For pollen that cannot be identified or is identified incorrectly, it is manually corrected and then input into the model training dataset.

[0029] By adopting the above technical solution, pollen that cannot be identified or is identified incorrectly can be manually corrected and then input into the model training dataset. This expands and corrects the model training dataset, thereby continuously improving the recognition rate and accuracy of the AI ​​data analysis model in practice.

[0030] Optionally, in step S2, the minimum distance between the upper and lower cutting positions and the minimum distance between the cutting position and the upper and lower pollen are set in the program beforehand, and the program automatically determines the cutting position through this logic.

[0031] By adopting the above technical solution, the cutting position can be automatically determined, avoiding cutting pollen.

[0032] In summary, this application includes at least one of the following beneficial technical effects:

[0033] 1. By using agar as a carrier for pollen, pollen is less likely to be layered on the same plane. The pollen is in a three-dimensional suspended state in the agar block, which avoids the flattening and deformation of the pollen and makes it easier to identify.

[0034] 2. Each captured image does not contain pollen from the lower layer of the agar block, and the pollen in the captured images is less likely to overlap, thus greatly reducing the occurrence of pollen overlap and improving the recognition rate;

[0035] 3. Improve recognition speed and accuracy through AI data analysis models. Attached Figure Description

[0036] Figure 1 This is a front view of Embodiment 1;

[0037] Figure 2 This is a partial top view of Embodiment 1;

[0038] Figure 3 This is a partial perspective view of Embodiment 1;

[0039] Figure 4 This is a schematic diagram illustrating the principle of determining the cutting position in the embodiment;

[0040] Figure 5This is a schematic diagram of the steps in an embodiment.

[0041] Explanation of reference numerals in the attached diagram: 1. Camera 1; 2. Camera 2; 3. Lifting platform; 4. Agar block; 5. Placement platform; 6. Conveyor belt mechanism; 31. Slide plate; 32. Linear module 1; 33. Linear module 2; 21. Three-dimensional moving platform; 22. Top light; 11. Side light; 34. Cutter; 41. Agar layer; 35. Baffle 1; 51. Linear module 3; 36. Electric cylinder; 37. Baffle 2. Detailed Implementation

[0042] The present application will be further described in detail below with reference to the accompanying drawings.

[0043] Example 1:

[0044] Reference Figure 1 and Figure 2 This embodiment discloses an artificial intelligence identification and analysis system for airborne pollen, including a PC, a frame (not fully shown in the figure), a camera 1, a camera 2, a lifting seat 3, an agar block containing pollen 4, a placement platform 5 for placing the agar block 4, and a conveyor belt mechanism 6. A sliding plate 31 is horizontally slidably connected to the lifting seat 3. The lifting seat 3 is driven to rise and fall by a linear module 32, and the sliding plate 31 is driven to slide by a linear module 33.

[0045] Camera 1 is located outside the side wall of agar block 4 and horizontally facing agar block 4. Camera 2 is located above agar block 4 and vertically facing agar block 4. A three-dimensional moving platform 21 is mounted on the frame. Camera 2 is fixed to the movable end of the three-dimensional moving platform 21, which is an XYZ three-axis moving platform composed of three servo motor lead screw mechanisms, enabling precise adjustment of the spatial position of camera 2. A top light 22 is fixed next to camera 2, facing agar block 4, illuminating the top surface of agar block 4 and providing supplemental lighting for camera 2. Camera 2 is used for image recognition, and a high-resolution camera or microscope camera can be selected. Camera 2 has a built-in filter to filter reflections.

[0046] Camera 1 is fixed to the frame and is used to capture images of pollen locations. A high-resolution camera is selected. Camera 1 is located outside the side wall of agar block 4 and horizontally facing agar block 4. A side lamp 11 is fixed to the frame and is located on the side of agar block 4 away from camera 1, horizontally facing agar block 4. Specifically, in this embodiment, camera 1 is located on the front side of agar block 4, and side lamp 11 is located on the rear side of agar block 4. Through the side lamp 11, camera 1 is located on the back side of agar block 4, and camera 1 can clearly capture images of pollen distribution within agar block 4.

[0047] The signal output terminals of Camera 1 and Camera 2 are connected to the PC. The PC controls the operation of Linear Module 1 32 and Linear Module 2 33 through the controller. Both Linear Module 1 32 and Linear Module 2 33 are driven by servo motors and have the feature of precise displacement adjustment.

[0048] Reference Figure 1 and Figure 3 A cutter 34 is fixed to the lower end of the slide plate 31. The cutter 34 is horizontally positioned, with its blade located at the end furthest from the slide plate 31. The cutter 34 is used to horizontally cut the agar block 4. By controlling the linear module 33, the cutter 34 cuts the agar block 4 in two strokes. The first stroke is immediately after the cut is completed, and the cut agar layer 41 is still on the cutter 34. The second stroke is as the cutter 34 continues to move, the slide plate 31 pushes the cut agar layer 41 to separate it from the agar block 4 below.

[0049] The lifting platform 3 is fixed with a baffle 35, which is vertically positioned to contact the side wall of the agar block 4. The top of the baffle 35 corresponds to the bottom of the cutter 34 at the same height. When the cutter 34 cuts the agar block 4, the baffle 35 supports the right side of the agar block 4, ensuring that the cutter 34 can cut the upper layer of the agar block 4 while the lower layer of the agar block 4 remains stationary. The height of the baffle 35 rises and falls with the lifting platform 3, providing optimal blocking for the lower layer of the agar block 4.

[0050] The placement platform 5 is slidably connected to the frame. The placement platform 5 is driven to slide by the linear module 3 51. After the placement platform 5 moves to the outer side with its top unobstructed, it is convenient to place the agar block 4. After the placement platform 5 slides towards the baffle 35, the side wall of the agar block 4 abuts against the baffle 35, and then subsequent steps can be performed.

[0051] The conveyor belt mechanism 6 is located on the side of the agar block 4 away from the slide plate 31, and the conveyor belt mechanism 6 is used to place the cut agar layer 41. The lifting seat 3 is equipped with an electric cylinder 36 and a baffle 37 driven by the electric cylinder 36 to lift. The electric cylinder 36 is fixed to the right end of the lifting seat 3.

[0052] Reference Figure 3 and Figure 5The specific operating conditions of the electric cylinder 36 and the second baffle 37 are as follows: When the slide plate 31 slides to the right limit, the agar layer 41 on the cutter 34 is directly above the conveyor belt mechanism 6. Then, the electric cylinder 36 drives the second baffle 37 to move downward. The second baffle 37 is used to insert between the slide plate 31 and the agar layer 41. Then, the slide plate 31 moves to the left and resets. Through the obstruction of the second baffle 37, the agar layer 41 loses the support of the cutter 34 and falls downward onto the conveyor belt mechanism 6. The conveyor belt mechanism 6 conveys each agar layer 41 backward a certain distance. The agar layers 41 are arranged on the conveyor belt mechanism 6, which facilitates subsequent management and the need for reuse of the agar layers 41.

[0053] The usage method of the system structure in this embodiment is described in Embodiment 2.

[0054] Example 2:

[0055] An artificial intelligence-based method for identifying and analyzing airborne pollen, using an artificial intelligence-based system for identifying and analyzing airborne pollen as described in Example 1, includes the following steps:

[0056] Step S1: Prepare agar blocks 4 containing pollen using a mold. The pollen originates from a glass slide, which is pre-captured from the air using Vaseline and agar. The pollen is then stained by the agar. When preparing agar blocks 4, a transparent agar material is used. While the agar is still liquid, the material on the top surface of the glass slide is poured into the mold and stirred, allowing the pollen to freely separate into layers within the agar. After the agar solidifies, agar blocks 4 are formed. By using agar as a carrier for the pollen, the pollen is less likely to layer on the same plane, and it maintains its original three-dimensional shape, making it easy to identify.

[0057] Step S2: Place the agar block 4 on the placement platform 5, so that one side of the agar block 4 is in close contact with the baffle 35. Turn on the side lamp 11 and the camera 1. The camera 1 takes an image of the pollen distribution inside the agar block 4. The image is used by the image recognition software in the PC to determine the position of multiple horizontal cuts, and the cuts avoid the pollen.

[0058] Reference Figure 4 The logic for determining the cutting position is as follows: First, the minimum distance between the upper and lower cutting positions is set in the program (e.g., 10mm), and the minimum distance between the cutting position and the upper and lower pollen grains (e.g., 3mm). The program automatically determines the cutting position based on this logic. Specifically, the cutting line is scanned from top to bottom. Every 10mm, the distance between the scan line and the nearest upper and lower pollen grains is calculated. If there are pollen grains within 3mm above and below the scan line, then the position of the scan line is taken as a cutting position.

[0059] Step S3: Refer to Figure 1 and Figure 5Based on the horizontal cutting height, the height of the lifting seat 3 is adjusted by controlling the operation of the linear module 1 32 until the cutter 34 reaches the corresponding cutting position. Then, the linear module 2 33 works, and the cutter 34 moves horizontally to cut the upper layer of the agar block 4 into an agar layer 41. After the cutting is completed, the operation is paused. At this time, the cut agar layer 41 is located on the cutter 34. The lifting seat 3 moves upward to bring the agar layer 41 closer to the camera 2. Through the separation of the cutter 34, the camera 2 will not capture the pollen in the lower layer.

[0060] Step S4: Camera 2 captures an image of agar layer 41 and transmits it to the PC. Using image recognition software combined with an AI data analysis model, pollen is identified, classified, and statistically analyzed. When Camera 2 is a high-resolution camera, the image is directly captured and transmitted to the PC. When Camera 2 is a microscope camera, it is moved by a three-dimensional moving platform 21. Camera 2 scans and captures the image, which is then stitched together on the PC to form a complete image.

[0061] The AI ​​data analysis model includes a model training dataset, which has been pre-trained and has a high recognition rate. For pollen that cannot be recognized or is recognized incorrectly, it is manually corrected and then input into the model training dataset, thereby expanding and correcting the model training dataset, and thus continuously improving the recognition rate and accuracy of the AI ​​data analysis model in practice.

[0062] Since the images captured each time do not contain pollen below the agar layer 41, the pollen in the captured images is not easily layered, thus greatly reducing the situation of pollen layering and improving the recognition rate; since the pollen is in a three-dimensional suspended state in the agar layer 41, the situation of pollen collapsing and deforming is also avoided, which can also improve the recognition rate.

[0063] Step S5: Remove the agar layer 41 on the cutter 34. After the cutter 34 is reset, repeat steps S3-S4 to cut the next agar layer 41 and take images, thereby completing the image taking and identification of all pollen.

[0064] The removal of the agar layer 41 on the cutter 34 is achieved by the translation of the slide plate 31 and the coordinated action of the electric cylinder 36 and the baffle 37. The conveyor belt mechanism 6 transports each agar layer 41 backward a certain distance, and the agar layers 41 are arranged on the conveyor belt mechanism 6 to facilitate subsequent management and the need for reuse of the agar layers 41.

[0065] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. An artificial intelligence-based method for identifying and analyzing airborne pollen, characterized in that: An artificial intelligence identification and analysis system for airborne pollen is used. The system includes a PC, a frame, a camera 1 (1), a camera 2 (2), a lifting seat (3), an agar block (4) containing pollen, and a placement platform (5) for placing the agar block (4). A sliding plate (31) is horizontally connected to the lifting seat (3). The lifting seat (3) is driven to rise and fall by a linear module 1 (32). The sliding plate (31) is driven to slide by a linear module 2 (33). A cutter (34) is fixed at the lower end of the sliding plate (31). The cutter (34) is horizontally set. The camera 1 (1) is located outside the side wall of the agar block (4) and horizontally facing the agar block (4). The camera 2 (2) is located above the agar block (4) and vertically facing the agar block (4). The signal output terminals of the camera 1 (1) and camera 2 (2) are connected to the PC. The PC controls the operation of the linear module 1 (32) and the linear module 2 (33) through a controller. The method includes the following steps: Step S1: Make agar blocks containing pollen using a mold (4); Step S2: Place the agar block (4) on the placement platform (5), start the camera (1), and the camera (1) takes an image of the pollen distribution in the agar block (4). The image is used by the image recognition software in the PC to determine the position of multiple horizontal cuts, and the cuts avoid the pollen. Step S3: Based on the height position of the horizontal cut, by controlling the operation of the linear module one (32), adjust the height position of the lifting seat (3) until the cutter (34) reaches the corresponding cutting position. Then the linear module two (33) works, and the cutter (34) moves horizontally to cut the upper layer of the agar block (4) into an agar layer (41). After the cutting is completed, pause and the lifting seat (3) moves up to bring the agar layer (41) closer to the camera two (2). Step S4: Camera 2 (2) captures an image of the agar layer (41) and sends it to the PC. The pollen is identified, classified and counted by image recognition software combined with AI data analysis model. Step S5: Remove the agar layer (41) on the cutter (34), and repeat steps S3-S4 after the cutter (34) is reset.

2. The method for artificial intelligence identification and analysis of airborne pollen according to claim 1, characterized in that: The AI ​​data analysis model includes a model training dataset. For pollen that cannot be identified or is identified incorrectly, it is manually corrected and then input into the model training dataset.

3. The method for artificial intelligence identification and analysis of airborne pollen according to claim 1, characterized in that: In step S2, the logic for determining the cutting position is as follows: first, the minimum distance between the upper and lower cutting positions and the minimum distance between the cutting position and the upper and lower pollen are set in the program, and the program automatically determines the cutting position through this logic.

Images