An artificial intelligence-based microplastic length calculation method

Abstract

The application provides an artificial intelligence-based microplastic length calculation method, and relates to the technical field of microplastic length detection.The method comprises the following steps: classifying and detecting an obtained image of a target microplastic particle to obtain a morphology type of the target microplastic particle; inputting the image of the target microplastic particle and the morphology type into a trained key point detection convolutional neural network to obtain corresponding key points and coordinates corresponding to the number of the morphology type; and calculating the length of the target microplastic particle corresponding to the morphology type according to the key points and coordinates to obtain a calculation result.The application greatly improves the efficiency of microplastic length detection.

Description

Technical Field

[0001] This invention relates to the field of microplastic length detection technology, and in particular to an artificial intelligence-based method for calculating the length of microplastics. Background Technology

[0002] Microplastics are plastic fragments or particles with a diameter of less than 5 millimeters, commonly referred to as "microplastics." As a new type of environmental pollutant, microplastics have been present in the environment for decades. For example, cleaning sponges are widely used as common cleaning tools in daily life. During use, sponges inevitably release microplastic particles due to friction. These plastic particles can have serious impacts on the environment and human health, polluting rivers, lakes, oceans, and other freshwater and saltwater environments. They can also contaminate human health; some studies even suggest that cleaning sponges, once ingested, may affect the nervous and cardiovascular systems. Therefore, effective measurement and assessment of microplastics are necessary.

[0003] Against this backdrop, calculating the length of microplastics not only aids in environmental monitoring to understand the extent and scope of microplastic pollution, but also helps identify pollution sources. It is a crucial component in researching and managing microplastic pollution, contributing to a more comprehensive understanding, monitoring, and response to the impacts of microplastics on the environment and ecosystems. Currently, the determination of microplastic length primarily relies on microscopic observation, estimating length through manual observation. This method is not only subject to subjective errors but also highly inefficient. Summary of the Invention

[0004] To overcome the shortcomings of existing technologies, the purpose of this invention is to provide a method for calculating the length of microplastics based on artificial intelligence.

[0005] To achieve the above objectives, the present invention provides the following solution:

[0006] An artificial intelligence-based method for calculating the length of microplastics includes:

[0007] The acquired images of the target microplastic particles are classified and detected to obtain the morphological type of the target microplastic particles;

[0008] The image of the target microplastic particles and the morphology type are input into a trained keypoint detection convolutional neural network to obtain the corresponding number of keypoints and coordinates for the morphology type.

[0009] The length of the target microplastic particle corresponding to the morphological type is calculated based on the key points and coordinates to obtain the calculation result.

[0010] Preferably, the morphological types include a first type, a second type, and a third type; the first type is a linear structure and the image has no crosslinking points; the second type is a branched structure and the image has one crosslinking point; the third type is a branched structure and the image has two crosslinking points.

[0011] Preferably, the image of the target microplastic particles and the morphology type are input into a trained keypoint detection convolutional neural network to obtain a corresponding number of keypoints and coordinates for the morphology type, including:

[0012] The image of the target microplastic particles of the first type is input into a trained key point detection convolutional neural network to obtain three key points and their coordinates;

[0013] The image of the target microplastic particles of the second type is input into a trained key point detection convolutional neural network to obtain four key points and their coordinates;

[0014] The image of the target microplastic particles of the third type is input into a trained keypoint detection convolutional neural network to obtain six keypoints and their coordinates.

[0015] Preferably, the length of the target microplastic particle corresponding to the morphological type is calculated based on the key points and coordinates to obtain the calculation result, including:

[0016] The key points corresponding to the first type are divided into a first center point P. m1 (x m1 y m1 ) and the two first boundary endpoints P 11 (x 11 y 11 ), P 21 (x 21 y 21 And connect the first center point and the two first boundary endpoints to obtain the first straight line;

[0017] According to the formula Calculate the length of the first straight line; where l1 is the length of the first straight line;

[0018] The calculation result is obtained by multiplying the length of the first straight line by the length of each pixel; or

[0019] The key points corresponding to the second type are divided into a second center point P. m2 (x m2 y m2 ) and three second boundary endpoints P 12 (x 12 y12 ), P 22 (x 22 y 22 ), P 32 (x 32 y 32 And connect the second center point to the three endpoints of the second boundary respectively to obtain the second line, the third line and the fourth line;

[0020] According to the formula Calculate the length of the second straight line according to the formula. Calculate the length of the third straight line according to the formula. Calculate the length of the fourth line; where l2 is the length of the second line, l3 is the length of the third line, and l4 is the length of the fourth line;

[0021] Add the lengths of the two longest lines among the second, third, and fourth lines, and multiply the sum by the length of each pixel to obtain the calculation result; or

[0022] The key points corresponding to the third type are divided into two third center points P. m3 (x m3 y m3 ), P m3′ (x m3′ y m3′ ) and four third boundary endpoints P 13 (x 13 y 13 ), P 23 (x 23 y 23 ), P 33 (x 33 y 33 ), P 43 (x 43 y 43 The second line is formed by connecting the two third center points to obtain the fifth line. A third center point is then connected to two third boundary endpoints to obtain the sixth and seventh lines. The third center point is then connected to two other third boundary endpoints to obtain the eighth and ninth lines. The third boundary endpoint P... 13 (x 13 y 13 ), P 23 (x 23 y 23 () is the closest point to the third center point P m3 (x m3 y m3 The two endpoints of ); the third boundary endpoint P 33 (x33 y 33 ), P 43 (x 43 y 43 () is the closest point to the third center point P m3′ (x m3′ y m3′ The two endpoints of );

[0023] According to the formula Calculate the length of the fifth straight line according to the formula. Calculate the length of the sixth line according to the formula. Calculate the length of the seventh line; according to the formula Calculate the length of the sixth line according to the formula. Calculate the length of the seventh line; where l5 is the length of the fifth line, l6 is the length of the sixth line, l7 is the length of the seventh line, l8 is the length of the eighth line, and l9 is the length of the ninth line;

[0024] The lengths of the longest lines among the sixth and seventh lines, the eighth and ninth lines, and the fifth line are added together, and the result is multiplied by the length of each pixel to obtain the calculation result.

[0025] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0026] This invention provides an artificial intelligence-based method for calculating the length of microplastics, comprising: classifying and detecting an acquired image of a target microplastic particle to obtain its morphological type; inputting the image of the target microplastic particle and the morphological type into a trained keypoint detection convolutional neural network to obtain a corresponding number of keypoints and their coordinates for the morphological type; and calculating the length of the target microplastic particle corresponding to the morphological type based on the keypoints and coordinates to obtain the calculation result. This invention significantly improves the efficiency of microplastic length detection. Attached Figure Description

[0027] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0028] Figure 1A flowchart of the method provided in an embodiment of the present invention;

[0029] Figure 2 This is a schematic diagram of a first type of microplastic particles provided in an embodiment of the present invention;

[0030] Figure 3 This is a schematic diagram of a second type of microplastic particles provided in an embodiment of the present invention;

[0031] Figure 4 This is a schematic diagram of a third type of microplastic particles provided in an embodiment of the present invention;

[0032] Figure 5 This is a schematic diagram of the detection results of the first type of microplastic particles provided in an embodiment of the present invention;

[0033] Figure 6 This is a schematic diagram of the detection results of the second type of microplastic particles provided in an embodiment of the present invention;

[0034] Figure 7 This is a schematic diagram of the detection results of the third type of microplastic particles provided in an embodiment of the present invention. Detailed Implementation

[0035] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0036] The purpose of this invention is to provide an artificial intelligence-based method for calculating the length of microplastics, which can greatly improve the efficiency of microplastic length detection.

[0037] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0038] Figure 1 The method flowchart provided in the embodiments of the present invention is as follows: Figure 1 As shown, this invention provides a method for calculating the length of microplastics based on artificial intelligence, including:

[0039] Step 100: Classify and detect the acquired images of the target microplastic particles to obtain the morphological type of the target microplastic particles;

[0040] Step 200: Input the image of the target microplastic particles and the morphology type into the trained keypoint detection convolutional neural network to obtain the corresponding number of keypoints and coordinates for the morphology type;

[0041] Step 300: Calculate the length of the target microplastic particle corresponding to the morphological type based on the key points and coordinates, and obtain the calculation result.

[0042] Preferably, the morphological types include a first type, a second type, and a third type; the first type is a linear structure and the image has no crosslinking points; the second type is a branched structure and the image has one crosslinking point; the third type is a branched structure and the image has two crosslinking points.

[0043] Specifically, this invention uses the YOLOv8 deep learning model as a baseline. The model consists of two stages. The first stage classifies the target and calculates the number of particles in each category. After classification, keypoint detection is performed for each category. First, the captured microplastic particles are classified and detected. In this invention, there are three types of microplastic particles extracted from three types of domestic household cleaning sponges. Based on their morphological characteristics, they are named type I (linear structure, no crosslinking points), type II (branched structure, one crosslinking point), and type III (branched structure, two crosslinking points), with morphologies as shown below. Figures 2 to 4 As shown.

[0044] Preferably, the image of the target microplastic particles and the morphology type are input into a trained keypoint detection convolutional neural network to obtain a corresponding number of keypoints and coordinates for the morphology type, including:

[0045] The image of the target microplastic particles of the first type is input into a trained key point detection convolutional neural network to obtain three key points and their coordinates;

[0046] The image of the target microplastic particles of the second type is input into a trained key point detection convolutional neural network to obtain four key points and their coordinates;

[0047] The image of the target microplastic particles of the third type is input into a trained keypoint detection convolutional neural network to obtain six keypoints and their coordinates.

[0048] After classification, keypoint detection is performed for each type of target. Specifically, based on the first-stage detection, the image is input into a pre-trained keypoint detection convolutional neural network (the pose estimation network attached to the YOLOv8 model). For each target detected in the first stage, the corresponding number of keypoints and coordinates are output according to its category as preparation data for calculating the length of the target (this invention calculates the length of microplastics by fitting a segmented polyline curve to the actual target length curve. Therefore, part of the keypoints includes the prominent endpoints of the target, which serve as the start and end points of the fitted curve, while the other part consists of the midpoints of the polyline during the fitting process, which are the places with greater curvature in the actual curve).

[0049] Preferably, the length of the target microplastic particle corresponding to the morphological type is calculated based on the key points and coordinates to obtain the calculation result, including:

[0050] The key points corresponding to the first type are divided into a first center point P. m1 (x m1 y m1 ) and the two first boundary endpoints P 11 (x 11 y 11 ), P 21 (x 21 y 21 And connect the first center point and the two first boundary endpoints to obtain the first straight line;

[0051] According to the formula Calculate the length of the first straight line; where l1 is the length of the first straight line;

[0052] The calculation result is obtained by multiplying the length of the first straight line by the length of each pixel; or

[0053] The key points corresponding to the second type are divided into a second center point P. m2 (x m2 y m2 ) and three second boundary endpoints P 12 (x 12 y 12 ), P 22 (x 22 y 22 ), P 32 (x 32 y 32 And connect the second center point to the three endpoints of the second boundary respectively to obtain the second line, the third line and the fourth line;

[0054] According to the formula Calculate the length of the second straight line according to the formula. Calculate the length of the third straight line according to the formula. Calculate the length of the fourth line; where l2 is the length of the second line, l3 is the length of the third line, and l4 is the length of the fourth line;

[0055] Add the lengths of the two longest lines among the second, third, and fourth lines, and multiply the sum by the length of each pixel to obtain the calculation result; or

[0056] The key points corresponding to the third type are divided into two third center points P. m3 (x m3 y m3 ), P m3′ (x m3′ y m3′ ) and four third boundary endpoints P 13 (x 13 y 13 ), P 23 (x 23 y 23 ), P 33 (x 33 y 33 ), P 43 (x 43 y 43 The second line is formed by connecting the two third center points to obtain the fifth line. A third center point is then connected to two third boundary endpoints to obtain the sixth and seventh lines. The third center point is then connected to two other third boundary endpoints to obtain the eighth and ninth lines. The third boundary endpoint P... 13 (x 13 y 13 ), P 23 (x 23 y 23 () is the closest point to the third center point P m3 (x m3 y m3 The two endpoints of ); the third boundary endpoint P 33 (x 33 y 33 ), P 43 (x 43 y 43 () is the closest point to the third center point P m3′ (x m3′ y m3′ The two endpoints of );

[0057] According to the formula Calculate the length of the fifth straight line according to the formula. Calculate the length of the sixth line according to the formula. Calculate the length of the seventh line; according to the formula Calculate the length of the sixth line according to the formula. Calculate the length of the seventh line; where l5 is the length of the fifth line, l6 is the length of the sixth line, l7 is the length of the seventh line, l8 is the length of the eighth line, and l9 is the length of the ninth line;

[0058] The lengths of the longest lines among the sixth and seventh lines, the eighth and ninth lines, and the fifth line are added together, and the result is multiplied by the length of each pixel to obtain the calculation result.

[0059] In this embodiment, for type I, there are three key points, namely, a center point P of the target. m (x m y m The center point is connected to two boundary endpoints P1(x1, y1) and P2(x2, y2). A straight line is formed by connecting the center point to the two boundary endpoints. The length of the straight line is calculated, and then multiplied by the length of each pixel to obtain the actual length of the target. For type II, there are four key points in total, namely the center point P1(x1, y1) and P2(x2, y2). m (x m y m And three boundary endpoints P1(x1, y1), P2(x2, y2), P3(x3, y3) Calculate P m Distance from P1 P m Distance from P2 P m Distance from P3 Compare the magnitudes of l1, l2, and l3, and discard the distance P. m The shortest point (let's assume P3) has a target length l = l1 + l2. Multiplying this length by the length of each pixel gives the actual target length. For type III, there are six keypoints in total, namely the two center points P. m1 (x m1 y m1 ), P m2 (x m2 y m2 ) and four boundary endpoints P 11 (x 11 y 11 ), P12 (x 12 y 12 ), P 21 (x 21 y 21 ) and P 22 (x 22 y 22 Of these four boundary endpoints, P 11 and P 12 Near P m1 P 21 and P 22 Near P m2 Based on this, first calculate P m1 With P m2 distance Calculate P again 11 and P 12 to P respectively m1 distance P 21 and P 22 to P respectively m2 distance Compare l respectively 11 and l 12 l 21 and l 22 The size, discarding the distance P m1 and P m2 Two smaller points (let's say P) 12 and P 22 If the length of the target is l = l, then the length of the target is l = l m +l 11 +l 21 The actual length of the target is then obtained by multiplying the length by the length of each pixel. Specific detection results are as follows: Figures 5 to 7 As shown.

[0060] Furthermore, in this invention, the length of the microplastic is calculated using a broken line fitting curve method. Strictly speaking, the center point is actually an endpoint of each fitted line segment. Training data is constructed by pre-marking the center point and boundary endpoints of each target on the microplastic image to train the key point detection network. This invention is still applicable to more complex types of microplastics. Therefore, this embodiment will not elaborate too much on the types of microplastic particles.

[0061] Optionally, in this embodiment, only new target detection and key point data need to be constructed based on the corresponding target structure. Microplastic length data has important applications in environmental monitoring; for example, the length of microplastics can be used to trace potential pollution sources. A specific application example is when urban drinking water sources are contaminated with microplastics. To determine the main pollution source, water samples from different drinking water sources, including rivers, lakes, or reservoirs, can be collected. The microplastic pollution levels in different areas can be compared. After obtaining the water source, the quantity and length of microplastics within the water source can be measured to obtain statistical information on the length distribution. Using artificial intelligence methods can greatly improve efficiency here. By comparing the length distribution data from different sampling points, such as the dominant length pattern, some sampling points may show one or more microplastics of specific lengths dominating, which may indicate that a certain pollution source caused this length pattern. For example, if the microplastic particles at a certain sampling point are generally short, this may be related to a specific type of plastic product or pollution source. Another example is the shape of the length distribution; the length distribution at some sampling points may show a peaked or skewed pattern, while other sampling points may show an average or multi-peaked distribution. This shape information can also provide clues about the pollution source. Next, emission estimates are conducted for suspected pollution sources. Combining the results of on-site investigations and source emission estimates, the primary pollution source is identified. If the emissions from a particular industrial plant or farm match the length distribution of microplastics found at the sampling points, then this source is likely the primary pollution source. Based on the identified pollution sources, governments and environmental agencies can take measures, such as improving wastewater treatment and educating factory managers to reduce plastic use, to reduce microplastic pollution and protect the quality of drinking water sources.

[0062] In this embodiment, three types of microplastics were identified and their lengths calculated. For other types of microplastic structures, the method described in this paper can still be used to calculate the length. Specifically, based on the target's shape, key points are determined. These key points are generally divided into two categories: one is the target's prominent endpoints (the beginning and end points of the target's true length curve), which serve as the starting and ending points for length calculation; the other is points with greater curvature (and a greater degree of bending) in the target's true length curve, which serve as the midpoints of the broken line in the fitted true curve. Finally, the distances between these points are calculated sequentially and summed to obtain the final length.

[0063] This invention presents three calculation methods for microplastics with different morphologies. For other microplastic types with more complex morphologies, i.e., multiple center points and boundary endpoints, this invention also provides a general calculation method. Based on the microplastic morphology and curvature, if a certain type of microplastic has N center points and M boundary endpoints, firstly, the boundary endpoints are associated with their corresponding center points. The principle is that, compared with other center points, when a boundary endpoint is closest to a certain center point, then the boundary endpoint is matched with that center point. Then, all the center points are connected into a line, and the length L1 of the line segment is calculated. Next, the distances between the two center points at both ends of the line and their respective boundary endpoints are calculated. The boundary endpoints farthest from the center point among their respective boundary endpoints are selected, with distances L2 and L3 respectively. Finally, the length of the microplastic L = L1 + L2 + L3.

[0064] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0065] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for calculating the length of microplastics based on artificial intelligence, characterized in that, include: The acquired images of target microplastic particles are classified and detected to obtain the morphological types of the target microplastic particles; the morphological types include a first type, a second type, and a third type; the first type is a linear structure and the image has no cross-linking points; the second type is a branched structure and the image has one cross-linking point; the third type is a branched structure and the image has two cross-linking points. The image of the target microplastic particles and the morphology type are input into a trained keypoint detection convolutional neural network to obtain a corresponding number of keypoints and coordinates for the morphology type, including: The image of the target microplastic particles of the first type is input into a trained key point detection convolutional neural network to obtain three key points and their coordinates; The image of the target microplastic particles of the second type is input into a trained key point detection convolutional neural network to obtain four key points and their coordinates; The image of the target microplastic particles of the third type is input into a trained key point detection convolutional neural network to obtain six key points and their coordinates; The length of the target microplastic particle corresponding to the morphological type is calculated based on the key points and coordinates to obtain the calculation result.

2. The method for calculating the length of microplastics based on artificial intelligence according to claim 1, characterized in that, The length of the target microplastic particle corresponding to the morphological type is calculated based on the key points and coordinates, and the calculation results are obtained, including: The key points corresponding to the first type are divided into a first center point. and the two first boundary endpoints And connect the first center point and the two first boundary endpoints to obtain the first straight line; According to the formula Calculate the length of the first straight line; where, The length of the first straight line; The calculation result is obtained by multiplying the length of the first straight line by the length of each pixel; or The key points corresponding to the second type are divided into a second center point. and three second boundary endpoints And connect the second center point to the three second boundary endpoints respectively to obtain the second straight line, the third straight line and the fourth straight line; According to the formula Calculate the length of the second straight line according to the formula. Calculate the length of the third straight line according to the formula. Calculate the length of the fourth straight line; where, The length of the second straight line. The length of the third straight line. The length of the fourth straight line; Add the lengths of the two longest lines among the second, third, and fourth lines, and multiply the sum by the length of each pixel to obtain the calculation result; or The key points corresponding to the third type are divided into two third center points. and four third boundary endpoints Connecting the two third center points yields the fifth straight line. Connecting one third center point to each of the two third boundary endpoints yields the sixth and seventh straight lines. Connecting the other third center point to the other two third boundary endpoints yields the eighth and ninth straight lines. The third boundary endpoints... The closest to the third center point The two endpoints; the third boundary endpoint The closest to the third center point The two endpoints; According to the formula Calculate the length of the fifth straight line according to the formula. Calculate the length of the sixth line according to the formula. Calculate the length of the seventh line; according to the formula Calculate the length of the sixth line according to the formula. Calculate the length of the seventh line; where, The length of the fifth straight line. The length of the sixth straight line. The length of the seventh straight line. The length of the eighth straight line. The length of the ninth straight line; The lengths of the longest lines among the sixth and seventh lines, the eighth and ninth lines, and the fifth line are added together, and the result is multiplied by the length of each pixel to obtain the calculation result.

Images