An image recognition method and system for uniformity of a soil stabilization remediation traditional Chinese medicine agent

Abstract

The application discloses an image recognition method and system for uniformity of a soil stabilizing and remediation traditional Chinese medicine agent, and belongs to the technical field of soil remediation, and comprises the following steps: S1, fully stirring and mixing contaminated soil, agent particles and a tracer to obtain a sample to be measured; S2, collecting images of different sections of the sample to be measured, and extracting color features of the tracer in each section image to obtain corresponding color feature maps; S3, positioning centers of each tracer color feature in the color feature map, and performing inflation processing on the color feature of each tracer according to the particle size of the tracer, so that the color features of each tracer are of the same size; and S4, calculating the uniformity of the sample to be measured according to the color feature map of the tracer after the inflation processing. The application can improve the detection efficiency and detection accuracy of the mixing uniformity of soil and agents, and has important significance for the application of actual multiple-metal-polluted-site stabilizing and remediation engineering.

Description

Technical Field

[0001] This invention belongs to the field of soil remediation technology, and more specifically, relates to an image recognition method and system for the uniformity of agent mixing in soil stabilization and remediation. Background Technology

[0002] In recent years, heavy metal pollution in soils such as mining areas and farmland has become a significant problem. Due to the high toxicity and strong migration ability of heavy metals, they can enter the human body through the food chain, causing serious harm to human health. Therefore, effective treatment of soil contaminated with heavy metals is crucial. One of the most commonly used pollution remediation technologies is solidification and stabilization treatment. This technology involves adding remediation agents to the contaminated soil to passivate the heavy metals, gradually reducing their bioavailability, thereby achieving the goal of heavy metal soil remediation. It features short treatment time, wide applicability, and low cost.

[0003] In actual soil remediation operations, the effectiveness of mixing and breaking down the soil with the chemical agent is crucial. The main purpose of thorough mixing is to ensure sufficient contact and reaction between the agent and the pollutants. Therefore, the degree of mixing between the soil and agent particles has a significant impact on the overall remediation effect of the contaminated soil. Accurate identification of the soil-chemical mixing degree can provide a reference for improving the mixing degree, maximizing the effect of the agent, and thus achieving better soil remediation results.

[0004] Currently, there is a lack of relevant standards or methods for evaluating the degree of mixing between soil and remediation materials. In actual engineering projects, the judgment of the mixing effect of soil and agents mainly relies on the engineering experience of on-site technicians, resulting in low testing efficiency and accuracy. Methods for detecting mixing degree generally include: wet chemical analysis, differential scanning calorimetry, tomography, spectroscopy, and digital image processing. However, wet chemical analysis, differential scanning calorimetry, tomography, and spectroscopy all require testing instruments and cannot provide real-time feedback, making them unsuitable for industrial applications. Digital image processing technology processes images captured by image acquisition devices into data information. This data information can be collected by a computer based on the images, and then specialized image processing methods are applied to perform qualitative analysis of the desired components and calculate the overall target composition to assess its mixing degree. However, directly using digital image processing technology to evaluate the mixing degree between soil and remediation materials has the following problems:

[0005] (1) The soil pesticide particles have irregular shapes and are difficult to identify;

[0006] (2) Soil pesticide particles are small in size and difficult to distinguish with the naked eye. They are also numerous, making it difficult to distinguish pesticide particles from soil.

[0007] Therefore, directly using digital image processing technology to segment the characteristics of pesticide particles and soil will produce large image segmentation errors, resulting in low accuracy in detecting the uniformity of soil-pesticide mixing. Summary of the Invention

[0008] To address the shortcomings and improvement needs of existing technologies, this invention provides an image recognition method and system for the uniformity of soil-pesticide mixing in soil stabilization and remediation, aiming to improve the detection efficiency and accuracy of soil-pesticide mixing uniformity.

[0009] To achieve the above objectives, according to a first aspect of the present invention, an image recognition method for the uniformity of agent mixing in soil stabilization remediation is provided, comprising:

[0010] S1. After thoroughly mixing the contaminated soil, reagent particles, and tracer, a sample to be tested is obtained; wherein, the tracer and the reagent particles have the same movement trajectory and distribution pattern in the soil, and the tracer does not react with the mixture of contaminated soil and reagent particles;

[0011] S2. Acquire images of different cross sections of the sample to be tested, and extract the color features of the tracer in each cross section image to obtain the corresponding color feature map;

[0012] S3. Locate the center of each tracer color feature in the color feature map, and expand the color feature of each tracer from the center of the color feature according to the particle size of the tracer so that the color feature of each tracer is the same size;

[0013] S4. Calculate the homogeneity of the sample to be tested based on the color characteristic map of the tracer after the expansion treatment.

[0014] Furthermore, the tracer is colored quartz sand particles, and the particle size of the colored quartz sand particles is 10-20 mesh.

[0015] Furthermore, the contaminated soil includes sandy soil, sandy clay, clay, black soil, and red soil; the colored quartz sand particles are green and black; wherein, the correspondence between the contaminated soil and the color of the colored quartz sand particles is as follows: green quartz sand particles are added to sandy soil, sandy clay, black soil, and red soil; black quartz sand particles are added to clay.

[0016] Furthermore, the mass mixing ratio of the tracer to the contaminated soil is 1:1000-1:100.

[0017] Furthermore, in S3, the particle size of the tracer after dilation treatment is 40-50 pixels.

[0018] Furthermore, the particle size of the expanded tracer is 45 pixels.

[0019] Furthermore, S4 includes:

[0020] S41. Divide the color feature map of the expanded tracer into regions according to the actual size of the sample to be tested, and obtain the total number of regions M.

[0021] S42. Determine the number of effective mixing regions m in the total number of divided regions M; wherein the tracer characteristic value of the effective mixing region is within a set threshold range;

[0022] S43. Calculate the homogeneity D of the sample to be tested:

[0023] Furthermore, in S41, the side length of each divided region is 300-800 pixels.

[0024] Furthermore, in S42, the set threshold range is: in, Xi represents the area occupied by the tracer pixel in the i-th partitioned region.

[0025] According to a second aspect of the present invention, an image recognition system for the uniformity of agent mixing in soil stabilization remediation is provided, comprising a computer-readable storage medium and a processor;

[0026] The computer-readable storage medium is used to store executable instructions;

[0027] The processor is configured to read executable instructions stored in the computer-readable storage medium and execute the method described in any of the first aspects.

[0028] In summary, the above-described technical solutions conceived in this invention can achieve the following beneficial effects:

[0029] (1) The identification method of the present invention adds a tracer to the contaminated soil and the soil remediation agent. Since the tracer has a uniform particle size distribution, the movement trajectory and distribution pattern of the tracer and the agent particles in the soil are consistent, and the tracer does not react with the mixture. Therefore, the color feature of the tracer can be extracted instead of directly extracting the feature of the agent particles, avoiding the problem that the soil agent particles are irregularly shaped and difficult to identify. At the same time, before calculating the homogeneity of the sample to be tested based on the color feature of the tracer, it is considered that the obtained tracer color feature map is a two-dimensional planar feature extraction map. The mixing of tracers and soil occurs in three-dimensional space. When the tracer is submerged in the soil and only a portion of its spatial features are exposed, directly calculating the mixing degree using the feature extraction map obtained in the two-dimensional plane would lead to significant errors. Therefore, the color features of each tracer are expanded according to its particle size to ensure that each tracer has equal weight in the calculation. That is, once a tracer feature is identified, it is considered fully exposed and counted as a complete particle. Tracers submerged in the soil are considered to be completely spread on the surface, improving the accuracy and scientific rigor of the mixing degree calculation. Furthermore, the method of this invention allows digital image processing methods to be applied to the detection of soil-pesticide mixing degree, enabling real-time feedback, improving detection efficiency, and making it suitable for industrial applications.

[0030] (2) Furthermore, this invention found that the movement trajectories and distribution patterns of colored quartz sand particles and reagent particles in soil are roughly the same, and colored quartz sand has more obvious color characteristics and larger particle characteristics, which facilitates machine vision recognition. Therefore, using colored quartz sand as a tracer can further improve the accuracy of detection. When the particle size of colored quartz sand is 10-20 mesh, the dynamic characteristics of the colored quartz sand particles are closest to the dynamic characteristics of reagent particles in soil.

[0031] (3) Furthermore, based on the color matching scheme of soil type and tracer color of the present invention, experiments have shown that it can achieve high detection accuracy.

[0032] (4) Preferably, when the mass mixing ratio of the tracer and the soil is between 1:1000 and 1:100, experiments have shown that a high detection accuracy can be obtained at this mixing ratio.

[0033] (5) Preferably, when the expansion particle size is between 40 and 50 pixels, the problem of small soil agent particles and large number of particles that make it difficult to distinguish the agent particles from the soil can be avoided. At the same time, experiments have shown that a high detection accuracy can be achieved at this expansion particle size.

[0034] (6) As a further preferred option, the best results can be achieved when the expansion particle size is 45 pixels.

[0035] (7) Furthermore, the mixing degree method of the present invention can quantitatively determine the mixing degree of the extracted agent-soil mixed sample image, which is convenient for evaluating and guiding the crushing and mixing operation of contaminated soil and remediation agent.

[0036] (8) As a preferred option, experiments have shown that when the side length of each divided region is 300-800 pixels, it can avoid the situation where there are too many blank areas due to the small area of ​​a single region, resulting in a low degree of uniformity in the calculation, and avoid the situation where the area of ​​a single region is too large, resulting in too few divided regions, resulting in a high degree of uniformity in the calculation, or even 100%, thereby further improving the accuracy of detection.

[0037] (9) Preferably, the movement characteristics of the agent and tracer in the soil are modeled using a statistical model, and combined with experiments, the set threshold range is found to be: At this time, a relatively accurate degree of mixing can be obtained.

[0038] In summary, the method of this invention can quickly and stably detect the uniformity of crushing and mixing of stabilization remediation materials and soil particles, providing technical support for the optimized design of soil stabilization remediation equipment and processing procedures. It is of great significance for practical engineering applications in the stabilization remediation of sites contaminated with multiple metals. Attached Figure Description

[0039] Figure 1 This is a flowchart of the image recognition method for the uniformity of agent mixing in soil stabilization and remediation according to the present invention.

[0040] Figure 2 This is a flowchart illustrating the soil agent mixing uniformity identification process in an embodiment of the present invention.

[0041] Figure 3 This is a schematic diagram of the color feature center of each tracer in the color feature map identified in the embodiments of the present invention.

[0042] Figure 4 This is a color feature map of the tracer obtained in Example 1 of the present invention.

[0043] Figure 5 This is a color feature map of the tracer obtained in Example 2 of the present invention.

[0044] Figure 6 This is a color feature map of the tracer obtained in Example 14 of the present invention.

[0045] Figure 7 This is a color feature map of the tracer obtained in Example 15 of the present invention.

[0046] Figure 8This is a color feature map of the tracer obtained in Example 16 of the present invention.

[0047] Figure 9 This is a color feature map of the tracer obtained in Example 17 of the present invention.

[0048] Figure 10 This is a color feature map of the tracer obtained in Example 18 of the present invention.

[0049] Figure 11 This is a color feature map of the tracer obtained in Example 29 of the present invention.

[0050] Figure 12 This is a color feature map of the tracer obtained in Example 34 of the present invention. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0052] like Figure 1 , Figure 2 As shown, the image recognition method for the uniformity of agent mixing in soil stabilization remediation according to the present invention includes:

[0053] S1. The contaminated soil, soil remediation agent, and corresponding tracer are added to the soil remediation equipment and thoroughly mixed to obtain the sample to be tested. The tracer and agent particles move in the soil with the same trajectory and distribution pattern, and the tracer does not react with the mixture. The mixture is a mixture of contaminated soil and soil remediation agent. Specifically, the tracer will not impair the use and function of the final mixture throughout the entire testing process.

[0054] S2. Acquire images of different cross sections of the sample to be tested, and extract the color features of the tracer in each cross section image to obtain the corresponding color feature map; In this embodiment of the invention, the color features of the tracer are the HSV color features of the tracer;

[0055] S3. Locate the center of the color feature of each tracer in the color feature map, and expand the color feature of each tracer from the center of the color feature according to the particle size of the tracer to make the color feature of each tracer the same size, thereby automatically completing the complete particle.

[0056] S4. Calculate the mixing uniformity of the sample to be tested, that is, the mixing uniformity of the agent and the soil, based on the color characteristic map of the tracer after the expansion treatment.

[0057] Specifically, in S1, the tracer used in this embodiment of the invention is colored quartz sand particles; the movement trajectory and distribution pattern of colored quartz sand particles and reagent particles in the soil are roughly the same, and colored quartz sand has more obvious color features and larger particle features, which is convenient for machine vision recognition.

[0058] Preferably, the particle size of the colored quartz sand is 10-20 mesh, at which point the kinetic characteristics of the colored quartz sand particles are closest to those of the pesticide particles in the soil.

[0059] Preferably, different tracer colors are selected based on different soil colors and different remediation agent colors. The tracer color scheme designed in this invention is shown in Table 1 below:

[0060] Table 1 Color Scheme for Soil, Reagents, and Tracers

[0061]

[0062] In Table 1, " / " means "or".

[0063] Experiments have shown that the tracer color matching scheme designed in this invention can improve the accuracy of uniformity detection.

[0064] In conventional engineering, the mixing ratio of soil remediation agents to soil is around 1:1000. In this embodiment of the invention, the mass mixing ratio of tracer to soil is between 1:1000 and 1:100. Experiments have shown that higher detection accuracy can be obtained at this mixing ratio. Preferably, experiments have shown that the best results can be achieved when the mass mixing ratio of tracer to soil is 1:100.

[0065] In this embodiment of the invention, considering that the clay has a fine texture and is easily adhering to the tracer after being mixed with it, thus interfering with image recognition, a method of blowing a breeze or spraying water mist is used to reduce the soil particles adhering to the surface of the tracer, thereby reducing interference and improving the recognition of the tracer in the clay.

[0066] In this embodiment of the invention, in step S2, images of the horizontal, vertical, and oblique sections of the sample to be tested are acquired, ensuring that the image space does not contain any other interfering content, i.e., the photographs only contain the soil-pesticide mixture. The image resolution should be at least 1500×1500 pixels, preferably at least 3500×3500 pixels, ensuring clear and well-lit photographs. Typically, the acquired images are in the RGB color space. In this embodiment, the RGB color space images of the different sections of the sample to be tested are converted to the HSV color space, and the HSV color features of the tracer in the images are extracted. HSV is closer to people's perceptual experience of color than RGB; it can very intuitively express the hue, vibrancy, and brightness of a color, facilitating color comparison and enabling better extraction of the tracer's sample color features. Furthermore, the tracer color matching scheme designed according to this invention can ensure that the color space ranges of the tracer and soil are separated as much as possible, improving the accuracy of tracer color feature extraction and thus improving the accuracy of mixing uniformity detection.

[0067] In S2, when extracting the color features of the tracer, the image recognition may misjudge the black tracer and the shadows formed by large particles. Therefore, the black part of the tracer's color features can be extracted by adding an algorithm filter and a secondary discrimination method. Then, non-tracer features are removed based on the shape and edge of the black.

[0068] Meanwhile, since some tracers are only partially visible and difficult to distinguish from images, model training can be used to improve the accuracy of feature recognition. Specifically, after acquiring images of different cross-sections of the sample to be tested, the tracers are labeled on the images based on the actual samples. This is used to complement the image recognition results, thereby improving the image recognition process. Through training on a large number of difficult-to-identify samples and manual correction, the accuracy of feature recognition is improved, ultimately enhancing the accuracy of uniformity identification.

[0069] Specifically, in S2, the color features of the tracer in the image are extracted to obtain the corresponding color feature map, including:

[0070] S21. Assign the color features of the extracted tracer to a white background image of the same size as the cross-sectional image of the sample to be tested to obtain a preliminary processed feature image. At this point, the tracer information extracted from the soil can be obtained.

[0071] S22. Perform inverse binarization thresholding on the feature image to obtain the color feature map of the tracer. The formula for inverse binarization thresholding is as follows: for points in the feature image with gray values ​​greater than the threshold, set their gray values ​​to 0; for points with gray values ​​less than or equal to the threshold, set their gray values ​​to the set maximum values.

[0072] Specifically, in S3, before locating the center of the color feature of each tracer in the color feature map, a secondary processing of the image is performed: the color feature map of the tracer is eroded to obtain a smooth tracer color feature map.

[0073] In S3, considering that the obtained tracer color feature map is a two-dimensional planar feature extraction map, while the mixing of the tracer and soil is carried out in three-dimensional space, when the tracer is buried by the soil and only a part of the spatial features are exposed, if the mixing degree is directly calculated using the two-dimensional planar feature extraction map obtained at this time, it will cause a large error. This is because, although the tracer is only exposed in a corner on the plane, it can undoubtedly be regarded as mixed in three dimensions.

[0074] Therefore, the obtained tracer color feature map is subjected to dilation processing to ensure that each tracer has equal weight in the calculation process. That is, as long as the tracer feature is identified, it is considered to be fully exposed and counted as a complete particle. Tracers submerged in the soil are considered to be completely spread on the surface, thus improving the accuracy and scientific nature of the mixing uniformity calculation. Specifically, such as Figure 3 As shown, the center of the color feature of each tracer in the color feature map is determined based on the geometry of the tracer. Then, based on the particle size of the tracer, the color feature of each tracer is expanded to ensure that the size of each tracer's color feature is the same. At this point, the expanded tracer color feature map can more accurately reflect the characteristics of the tracer (agent) in the soil.

[0075] As a preferred option, experiments have shown that a particle size of 40-50 pixels (i.e., the particle size of the tracer) can achieve high detection accuracy; as a further preferred option, a particle size of 45 pixels can achieve the best results.

[0076] Specifically, S4 includes:

[0077] S41. Divide the color feature map of the expanded tracer into regions (i.e., divide it into cells) according to the actual size of the sample to be tested, and obtain the total number of regions M.

[0078] S42. Determine the number of effective mixing regions m in the total number of regions M; where an effective mixing region refers to a region where the tracer characteristic value is within a set threshold range.

[0079] S43. Calculate the homogeneity D of the sample to be tested:

[0080]

[0081] Specifically, in S41, in this embodiment of the invention, the size of each region is divided according to the size of the tracer and the pixel size of the collected sample image. Preferably, the pixel segmentation side length of each region (i.e., the side length of each divided region) is between 300 and 800 pixels. This avoids situations where a single region is too small, resulting in too many blank areas and thus a low calculated uniformity, and also avoids situations where a single region is too large, resulting in too few divided regions and thus a high calculated uniformity, even reaching 100%. Simultaneously, the density of the tracer feature distribution can also be considered when dividing the regions. Specifically, when dividing the regions, if a tracer feature is located at the boundary of multiple regions, then the tracer feature is assigned to the region with the largest area occupied by that tracer feature.

[0082] In S42, in this embodiment of the invention, the set threshold range is: in, Xi represents the area occupied by the tracer pixel in the i-th partitioned region, which can be used to characterize the tracer feature value of that region; if The region where the tracer characteristic value of the i-th region falls within the set threshold range is considered an effective mixing region. This set threshold range is determined through a statistical model.

[0083] Specifically, the proportion of tracer can be calculated for the color characteristics of each cross section, and then the average can be taken to obtain the final mixing status of the tracer in three-dimensional space.

[0084] The method of the present invention will be further described below with specific embodiments.

[0085] Example 1

[0086] Yellow sandy soil from the Potou area of ​​Henan Province was used, along with lime (white), a commonly used soil remediation agent, and a green tracer. The mixture was then stirred using a mechanical mixer.

[0087] After mixing, take photos according to the above steps to identify the soil reagent mixing uniformity. The parameters are set as follows: pixel segmentation side length 400 pixels, tracer particle size 45 pixels, tracer-to-soil mass mixing ratio 1:100, and quartz sand particle size 10-20 mesh. The color feature image of the tracer obtained after processing using the method of this invention is shown below. Figure 4 As shown in Table 2, the calculated sample homogeneity, the actual sample homogeneity, and the accuracy calculated by the method of the present invention are shown in Table 2.

[0088] Example 2

[0089] Unlike Example 1, in this example, the added soil remediation agent is fly ash (black). Other parameters are the same as in Example 1. The color characteristic map of the tracer obtained after processing by the method of this invention is shown below. Figure 5 As shown in Table 2, the calculated sample homogeneity, the actual sample homogeneity, and the accuracy calculated by the method of the present invention are shown in Table 2.

[0090] Example 3

[0091] Unlike Example 1, in this example, the pixel segmentation side length is 600 pixels, and other parameters are the same as in Example 1. The calculated sample homogeneity, the actual sample homogeneity, and the accuracy calculated by the method of the present invention are shown in Table 2.

[0092] Examples 4-5

[0093] Unlike Example 2, in Examples 4 and 5, the pixel segmentation side lengths are 600 pixels and 800 pixels, respectively. Other parameters are the same as in Example 2. The calculated sample homogeneity, the actual sample homogeneity, and the accuracy of the method of the present invention are shown in Table 2.

[0094] Comparative Examples 1-4

[0095] Unlike Example 1, in Comparative Examples 1-2, the pixel segmentation side lengths are 200 pixels and 1000 pixels, respectively. Other parameters are the same as in Example 1. The calculated sample homogeneity, the actual sample homogeneity, and the accuracy calculated by the method of the present invention are shown in Table 2.

[0096] Unlike Example 2, in Comparative Examples 3-4, the pixel segmentation side lengths are 200 pixels and 10 pixels, respectively. Other parameters are the same as in Example 2. The calculated sample homogeneity, the actual sample homogeneity, and the accuracy of the method of the present invention are shown in Table 2.

[0097] Examples 6-9

[0098] Unlike Example 1, in Examples 6-9, the mixing ratios of the tracer and soil were 1:1000, 1:800, 1:500, and 1:300, respectively. Other parameters were the same as in Example 1. The calculated sample homogeneity, the actual sample homogeneity, and the accuracy of the method of the present invention are shown in Table 2.

[0099] Examples 10-13

[0100] Unlike Example 2, in Examples 10-13, the mixing ratios of the tracer and soil were 1:1000, 1:800, 1:500, and 1:300, respectively. Other parameters were the same as in Example 2. The calculated sample homogeneity, the actual sample homogeneity, and the accuracy of the method of the present invention are shown in Table 2.

[0101] Comparative Examples 5-6

[0102] Unlike Example 2, in Comparative Examples 5-6, the mixing ratios of tracer and soil were 1:10 and 1:2000, respectively. Other parameters were the same as in Example 2. The calculated sample homogeneity, the actual sample homogeneity, and the accuracy calculated by the method of the present invention are shown in Table 2.

[0103] Table 2. Experimental conditions and results of different embodiments and comparative examples.

[0104]

[0105]

[0106] In Tables 2 and 3, the white remediation agent is commonly used lime, and the black remediation agent is commonly used fly ash; the tracer used in the embodiments of this invention is colored quartz sand; the mixing ratio in the table is the mass mixing ratio of the tracer and the soil; the segmented side length refers to the pixel side length of each segmented region; the calculated uniformity refers to the soil agent uniformity calculated by the method of this invention, and the actual uniformity refers to the actual uniformity of the sample.

[0107] Example 14

[0108] In this embodiment, heavy metal-contaminated clay (yellow) from Hunan Province was taken, and a black tracer was added. The mixture was then mechanically stirred. The soil was then photographed to identify the uniformity of the agent mixing. The pixel segmentation side length was 700 pixels, and other parameters were the same as in Example 2. The color feature map of the tracer obtained after processing using the method of this invention is shown below. Figure 6 As shown in Table 3, the calculated sample homogeneity, the actual sample homogeneity, and the accuracy of the method of the present invention are as follows.

[0109] Examples 15-16

[0110] In Examples 15 and 16, red soil from Honghe, Yunnan Province was used. In Example 15, lime (white), a commonly used soil remediation agent, was added to the soil, along with a green tracer. The pixel segmentation side length was 300 pixels, and the tracer particle size was 40 pixels. Other parameters were the same as in Example 1. After processing using the method of this invention, the color feature map of the tracer was obtained as shown below. Figure 7 As shown in Example 16, fly ash (black) and a green tracer were added to the soil. The pixel segmentation side length was 500 pixels, the tracer particle size was 45 pixels, and other parameters were the same as in Example 1. After processing by the method of the present invention, the color feature map of the tracer was obtained as shown in the figure. Figure 8 As shown in Table 3, the calculated sample homogeneity, the actual sample homogeneity, and the accuracy of the method of this invention are all presented.

[0111] Examples 17-22

[0112] In Example 17, Jilin black soil was taken, and lime (white), a commonly used soil remediation agent, was added, followed by a green tracer. The mixture was then stirred using a mechanical mixer. The pixel segmentation side length was 700 pixels, and the remaining parameters were the same as in Example 1. After processing using the method of this invention, the color feature map of the tracer was obtained as shown below. Figure 9 As shown in Table 3, the calculated sample homogeneity, the actual sample homogeneity, and the accuracy of the method of the present invention are as follows.

[0113] In Examples 18-19, unlike Example 17, the tracer particle sizes were set to 40 pixels and 50 pixels, respectively. In Examples 20-21, unlike Example 17, the added soil remediation agent was fly ash (black), and the tracer particle sizes were set to 45 pixels, 40 pixels, and 50 pixels, respectively; the remaining parameters were the same as in Example 17. In Example 18, the color feature map of the tracer obtained after processing by the method of the present invention is shown below. Figure 10 As shown in Table 3, the calculated sample homogeneity, the actual sample homogeneity, and the accuracy of the method of the present invention are as follows.

[0114] Comparative Examples 7-8

[0115] The difference between Comparative Example 7 and Example 17 is that the tracer particle size is 70 pixels. The difference between Comparative Example 8 and Example 17 is that the added soil remediation agent is fly ash (black), and the tracer particle size is 20 pixels. The calculated sample homogeneity, the actual sample homogeneity, and the accuracy calculated by the method of the present invention are shown in Table 3.

[0116] Examples 23-28

[0117] Examples 23-25 ​​differ from Example 17 in that the pixel segmentation side lengths are 400 pixels, 600 pixels, and 800 pixels, respectively; Examples 26-28 differ from Example 17 in that the added soil remediation agent is fly ash (black), and the pixel segmentation side lengths are 400 pixels, 600 pixels, and 800 pixels, respectively; the remaining parameters are the same as in Example 17. The calculated sample homogeneity, the actual sample homogeneity, and the accuracy calculated by the method of the present invention are shown in Table 3.

[0118] Comparative Examples 9-10

[0119] The difference between Comparative Example 9 and Example 17 is that the pixel segmentation side length is 200 pixels. The difference between Comparative Example 10 and Example 17 is that the added soil remediation agent is fly ash (black), and the pixel segmentation side length is 200 pixels. The calculated sample homogeneity, the actual sample homogeneity, and the accuracy calculated by the method of the present invention are shown in Table 3.

[0120] Table 3. Experimental conditions and results of different embodiments and comparative examples.

[0121]

[0122] In Table 3, in Example 14, in order to avoid the influence of clay on the characteristics of the extracted tracer, no remediation agent was added to the clay in this example. However, this does not affect the calculation of the soil agent's mixing degree by the mixing degree of the tracer and the soil. Correspondingly, the actual mixing degree referred to in this example refers to the actual mixing degree of the tracer and the soil.

[0123] Example 29

[0124] A black soil area contaminated with multiple metals was treated using an in-situ crushing and mixing integrated soil remediation device from a certain company for soil stabilization and remediation. White metal remediation agents and green tracers were simultaneously added to the device's dosing hopper, followed by crushing and mixing. After mixing, the soil was photographed according to the above steps to identify the uniformity of the agent mixing. The parameters set were: pixel segmentation side length of 400 pixels, tracer particle size of 45 pixels, tracer-to-soil mass mixing ratio of 1:100, and quartz sand particle size of 10-20 mesh. The color feature image of the tracer obtained after processing using the method of this invention is shown below. Figure 11 As shown in Table 4, the calculated sample homogeneity, the actual sample homogeneity, and the accuracy of the method of the present invention are as follows.

[0125] Examples 30-33

[0126] The difference between Examples 30-31 and Example 29 is that the tracer particle sizes are 40 pixels and 50 pixels, respectively; the difference between Examples 32-33 and Example 29 is that the pixel segmentation side lengths are 600 pixels and 800 pixels, respectively; the other parameters are the same as in Example 29. The calculated sample homogeneity, the actual sample homogeneity, and the accuracy calculated by the method of the present invention are shown in Table 4.

[0127] Comparative Examples 11-12

[0128] Unlike Example 29, in Comparative Examples 11-12, the tracer colors were black and white, respectively, while the other parameters were the same as in Example 29. The calculated sample homogeneity, the actual sample homogeneity, and the accuracy calculated by the method of the present invention are shown in Table 4.

[0129] Table 4. Experimental conditions and results of different embodiments and comparative examples.

[0130]

[0131] Example 34

[0132] In Wuhu City, Anhui Province, a polluted area of ​​yellow soil (sandy soil) was treated using a tracked self-propelled in-situ integrated soil remediation equipment. White metallic remediation agents and green tracers were simultaneously added to the dosing hopper, followed by crushing and mixing. After mixing, the soil was photographed according to the above steps to identify the uniformity of the agent mixing. The parameters were set as follows: pixel segmentation side length 400 pixels, tracer particle size 45 pixels, tracer-to-soil mass mixing ratio 1:100, and quartz sand particle size 10-20 mesh. The color feature image of the tracer obtained after processing using the method of this invention is shown below. Figure 12 As shown in Table 5, the calculated sample homogeneity, the actual sample homogeneity, and the accuracy of the method of the present invention are as follows.

[0133] Examples 35-38

[0134] Unlike Example 34, in Examples 35 and 36, the tracer particle size was set to 40 pixels and 50 pixels, respectively; in Examples 37 and 38, the pixel segmentation side length was 600 pixels and 800 pixels, respectively, and the other parameters were the same as in Example 24. The calculated sample homogeneity, the actual sample homogeneity, and the accuracy calculated by the method of the present invention are shown in Table 5.

[0135] Examples 39-40

[0136] Unlike Example 34, in Examples 35 and 36, the mass mixing ratio of tracer to soil was 1:200 and 1:500, respectively. The remaining parameters were the same as in Example 24. The calculated sample homogeneity, the actual sample homogeneity, and the accuracy calculated by the method of the present invention are shown in Table 5.

[0137] Comparative Examples 13-14

[0138] Unlike Example 34, in Comparative Examples 13-14, the particle sizes of the quartz sand were 30-50 mesh and 60-80 mesh, respectively. The other parameters were the same as in Example 24. The calculated sample homogeneity, the actual sample homogeneity, and the accuracy calculated by the method of the present invention are shown in Table 5.

[0139] Table 5. Experimental conditions and results of different embodiments and comparative examples.

[0140]

[0141]

[0142] As can be seen from the above embodiments, the image recognition method for the uniformity of agent mixing in soil stabilization remediation of the present invention can achieve high detection accuracy.

[0143] The identification method of this invention involves adding a tracer to the contaminated soil and the soil remediation agent. Because the tracer has a uniform particle size distribution, its movement trajectory and distribution pattern in the soil are consistent with those of the agent particles. Furthermore, the tracer does not react with the mixture. Therefore, extracting the color characteristics of the tracer can replace directly extracting the characteristics of the agent particles, avoiding the problem of irregularly shaped soil agent particles being difficult to identify. Simultaneously, before calculating the homogeneity of the sample based on the tracer's color characteristics, the method considers that the obtained tracer color feature map is a two-dimensional planar feature extraction map, while the tracer... The mixing of tracer with soil occurs in three-dimensional space. When the tracer is submerged in soil and only a portion of its spatial features are exposed, directly calculating the mixing degree using the feature extraction map obtained in the two-dimensional plane at this point would lead to significant errors. Therefore, the color features of each tracer are expanded according to its particle size to ensure that each tracer has equal weight in the calculation process. That is, as long as the tracer feature is identified, it is considered to be fully exposed and counted as a complete particle. Tracers submerged in soil are considered to be completely spread on the surface, improving the accuracy, scientific rigor, and detection efficiency of the mixing degree calculation. Furthermore, the method of this invention can provide real-time feedback and is suitable for industrial applications.

[0144] The present invention also provides an image recognition system for the uniformity of agent mixing in soil stabilization remediation, comprising a computer-readable storage medium and a processor; the computer-readable storage medium is used to store executable instructions; the processor is used to read the executable instructions stored in the computer-readable storage medium and execute the image recognition method for the uniformity of agent mixing in soil stabilization remediation in the above embodiments.

[0145] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An image recognition method for the uniformity of agent mixing in soil stabilization and remediation, characterized in that, include: S1. After thoroughly mixing the contaminated soil, reagent particles, and tracer, a sample to be tested is obtained; wherein, the tracer and the reagent particles have the same movement trajectory and distribution pattern in the soil, and the tracer does not react with the mixture of contaminated soil and reagent particles; S2. Acquire images of different cross sections of the sample to be tested, and extract the color features of the tracer in each cross section image to obtain the corresponding color feature map; S3. Locate the center of each tracer color feature in the color feature image, and perform dilation processing on the color feature of each tracer from the center of the color feature according to the particle size of the tracer, so that the color feature of each tracer is the same size, in order to correct the two-dimensional image recognition error caused by the tracer being partially buried by the soil in three-dimensional space, so that each identified tracer has equal weight in the calculation process. S4. Calculate the homogeneity of the sample to be tested based on the color characteristic map of the tracer after the expansion treatment; The contaminated soil includes sandy soil, sandy clay, clay, black soil, and red soil; the tracer is colored quartz sand particles; the colored quartz sand particles are green and black; wherein, the correspondence between the contaminated soil and the color of the colored quartz sand particles is as follows: green quartz sand particles are added to sandy soil, sandy clay, black soil, and red soil; black quartz sand particles are added to clay. The mass mixing ratio of the tracer to the contaminated soil is 1:1000-1:100; In S3, the particle size of the tracer after dilation treatment is 40-50 pixels; S4 includes: S41. Divide the color feature map of the expanded tracer into regions according to the actual size of the sample to be tested, and obtain the total number of regions. ; S42. Determine the total number of regions. Number of effective mixing regions The tracer characteristic value of the effective mixing region is within a set threshold range. S43. Calculate the homogeneity of the sample to be tested. : ; In S42, the set threshold range is: ;in, ; Indicates the first The tracer pixel occupies the first division region. The area of ​​each divided region.

2. The method according to claim 1, characterized in that, The colored quartz sand particles have a particle size of 10-20 mesh.

3. The method according to claim 1, characterized in that, The particle size of the expanded tracer is 45 pixels.

4. The method according to claim 1, characterized in that, In S41, the side length of each divided region is 300-800 pixels.

5. An image recognition system for the uniformity of agent mixing in soil stabilization and remediation, characterized in that, Includes computer-readable storage media and processors; The computer-readable storage medium is used to store executable instructions; The processor is configured to read executable instructions stored in the computer-readable storage medium and execute the method according to any one of claims 1-4.

Images