Image interpretation method for identifying contaminant transport in a variably saturated porous medium
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- JILIN UNIVERSITY
- Filing Date
- 2024-03-20
- Publication Date
- 2026-06-05
Smart Images
Figure CN118154559B_ABST
Abstract
Description
Technical fields:
[0001] This invention relates to an image interpretation method for identifying the migration of contaminants in variable-saturation porous media. Background technology:
[0002] With socio-economic development, the infiltration of site waste, excrement, and landfill leachate poses a potential risk to the safety of drinking water. As a variable saturation zone where hydrogeological states transition, pollutants in this zone occupy a critical position in the migration and propagation of the soil-groundwater system. Internally, it exhibits heterogeneous variable saturation characteristics with multiple interfaces coexisting. The seepage movement it undergoes differs from the seepage patterns of the unsaturated zone and the groundwater flow mechanism of the saturated zone. Due to its unique hydraulic characteristics and the influence of solid-liquid-gas interactions, the migration and transformation of pollutants in variable saturated porous media are extremely complex.
[0003] To address the aforementioned issues, existing technologies typically employ indoor sand tank and sand column experiments, along with numerical simulations, to study water flow and pollutant migration in variable-saturation porous media. These studies analyze the macroscopic changes in rise height, soil moisture, and soil-water potential under different conditions to determine the dissolution of pollutants. However, due to the difficulty in sampling the variable-saturation zone during indoor experiments, it is challenging to determine the spatial heterogeneity of water saturation distribution and its phase distribution characteristics within the porous media from a microscopic perspective. Furthermore, existing numerical simulations, based on the linear permeability law, fail to consider the abrupt phase transition regions caused by the spatial heterogeneity of water saturation in the variable-saturation zone. They also lack a detailed characterization of the variable-saturation zone's structural features and a quantitative expression of its interface composition, making it difficult to accurately identify the migration behavior of pollutants in variable-saturation porous media. Summary of the Invention:
[0004] This invention provides an image interpretation method for identifying pollutant migration in variable-saturation porous media. The method features a rationally designed structure that integrates precise characterization of the structural features and interfaces of variable-saturation porous media into a quantitative representation. This allows for accurate quantitative identification of pollutants in variable-saturation porous media. It can reveal the spatial heterogeneity of water saturation distribution and its phase distribution characteristics from a microscopic perspective, thereby accurately identifying the migration behavior of pollutants in variable-saturation porous media. It can achieve precise identification at the nanometer-level pixel level for any region and can also achieve rapid response to millions or more data features and results, balancing data processing efficiency and accuracy, and solving the problems existing in the prior art.
[0005] The technical solution adopted by the present invention to solve the above-mentioned technical problems is as follows:
[0006] An image interpretation method for identifying contaminant migration in variable-saturation porous media, the image interpretation method comprising the following steps:
[0007] S1 continuously acquires images of the formation process of the variable saturation zone and the migration process of pollutants, and performs color space conversion and grayscale processing;
[0008] S2, based on grayscale spatial information elements and color spatial information elements, performs image preprocessing, and adjusts the overall changes and local fine features of the variable saturation band formation process and pollutant migration process to clarify the range of variable saturation porous media and pollutant plume change areas;
[0009] S3 enhances effective image information on the formation of variable saturation bands and pollutant migration processes, and magnifies image details in near-boundary areas to improve brightness distortion;
[0010] S4. By utilizing the information gradient differences of continuous images, the structural features and interface composition of the variable saturation zone formation process and the behavior of pollutant migration process are characterized to obtain the initial image.
[0011] S5 performs a difference operation on the acquired process image and the initial image, converts the pixel values into real values, and combines the water saturation calculation formula and the pollution saturation calculation formula to quantitatively evaluate the representative elements of the variable saturation porous media and its pollution migration, accurately quantitatively identify the pollution in the variable saturation porous media, and accurately identify the migration behavior of pollutants in the variable saturation porous media.
[0012] The process of continuously acquiring images of the formation of the variable saturation zone and the migration of pollutants, followed by color space conversion and grayscale processing, includes the following steps:
[0013] S1.1, Adjust camera settings, exposure time and frame rate to change the camera's dynamic range;
[0014] S1.2, automatically capture continuous images of the variable saturation band formation process and pollutant migration process, and set and store the pixel format of the images in MONO8 and RGB formats.
[0015] Image preprocessing based on grayscale and color spatial information elements, adjusting for overall changes and local fine features in the formation process of the variable saturation band and the pollutant migration process, and clarifying the range of the variable saturation porous medium and the pollution plume variation area includes the following steps:
[0016] S2.1 performs morphological processing on the grayscale image, expanding or shrinking white pixel blocks, connecting adjacent white areas, and eliminating isolated white areas; adjusting the shape of the morphological structural unit to a rectangle, ellipse, or cross shape, adjusting the kernel width and height, and increasing the number of iterations, thereby extracting image components that are highly correlated with variable saturation porous media and contaminated areas.
[0017] S2.2, Set the corresponding thresholds to binarize the grayscale images of the formation of variable saturation bands and the pollution process at multiple brightness levels. In order to avoid setting the minimum threshold, the maximum threshold, the filter kernel height and width settings for image smoothness, and the threshold offset settings under the condition of obvious black and white differences, so as to clarify the small pixel differences between the variable saturation band, the pollution plume range boundary and the medium.
[0018] S2.3 Noise suppression is performed while preserving the inherent details of the variable saturation band and pollutant migration characteristics. Multiple filter types are used to adjust the height, width and number of filter kernels for the image, and the smoothing value in the corresponding direction is increased to improve the smoothness of the image. At the same time, the edge clarity range is improved by changing the edge threshold range, thereby improving the effectiveness and reliability of identifying the formation of the variable saturation band and the pollution process.
[0019] The binarization process includes hard thresholding, Gaussian binarization, and mean binarization.
[0020] Enhancing the effective image information of variable saturation band formation and pollutant migration processes, and magnifying image details in near-boundary regions to improve brightness distortion, includes the following steps:
[0021] S3.1, change the sharpening kernel size and sharpening coefficient to determine the local change area and provide sharpening intensity, combined with the adjustment coefficient to control the contrast to change the difference between the variable saturation band, the pollution migration area and the medium in the image;
[0022] S3.2, set gain and compensation to change the overall pixel brightness of the image to achieve brightness correction, and adjust the Gamma value to increase the proportion of image units in areas of light and dark difference, perform non-linear color editing on the image, increase the contrast between the variable saturation band, the pollution migration area and the medium, thereby magnifying details.
[0023] By utilizing the information gradient differences in consecutive images, the structural features and interface composition of the variable saturation band formation process, as well as the behavior of pollutant migration, are characterized to obtain the initial image. This process includes the following steps:
[0024] S4.1, phase interface identification is performed on the image of the variable saturation zone formation process. A machine learning image segmentation algorithm is introduced to segment the phase interface on the variable saturation zone profile. A limited number of manual annotations are used to train the classifier and automatically segment the remaining interface data. The segmentation results of the gas-solid interface and the liquid-solid interface are optimized to generate the region area and obtain the ratio of liquid phase area to gas phase area in the pores of the variable saturation porous medium.
[0025] S4.2, detect and extract the saturation band and pollution migration process edge points of the grayscale image. After filtering and boundary enhancement processing of the image, the vertex edge is detected and located. Based on the image gradient that is different between the extreme point and the surrounding nearby pixels, the grayscale change of the region is calculated to obtain the local extreme point, thereby adjusting the suppression threshold to determine the vertex edge.
[0026] S4.3 extracts brightness and color from the color image by performing saturation band and pollution migration process, calculates the relationship between image pixel values and true values, uses pixel values to represent the range of the region of interest, and sets the corresponding center point coordinates and outer and inner diameters. Then, through color channel conversion, sets a reasonable threshold to suppress target and background areas, selects RGB / HSL channels to measure the average, maximum and standard deviation of color and brightness, thereby confirming the numerical changes of multiple channels under different color spaces.
[0027] The formula for calculating the water saturation is:
[0028]
[0029] Where, S(x, y) is the water saturation at each pixel, and L s(x，y) L is the brightness value of this pixel. air L represents the brightness value when the pores of a porous medium are completely filled with air. saturated This represents the brightness value when the porous medium is fully saturated with water.
[0030] The formula for calculating pollution saturation is:
[0031]
[0032] Where F(x, y) is the contamination saturation of each pixel, and L f(x，y) L is the brightness value of a pixel. background L represents the background brightness value before any pollutants are introduced. f，0 This represents the maximum brightness value when the pollutants are introduced into the porous medium during the initial stage of the experiment.
[0033] The image interpretation system of the image interpretation method includes:
[0034] An image acquisition device, which is an external device, is placed at the front end of an indoor sand tank or sand column plexiglass plate. It is used to detect the area where the variable saturation zone forms and pollutants migrate. When the variable saturation zone begins to form and pollutants begin to migrate, the image is acquired in real time and stored in a memory.
[0035] A computer medium includes a memory, a processor, and a decoder; the memory stores continuous images acquired by an image acquisition device; the processor processes images of variable saturation band formation and pollutant migration processes; and the decoder calculates representative elements from the images of the variable saturation porous medium and its pollutant migration obtained by the processor.
[0036] This invention employs the aforementioned structure, continuously acquiring images of the variable saturation zone formation process and pollutant migration process, performing color space conversion and grayscale processing; preprocessing the images using grayscale and color space information elements to adjust the overall changes and local fine features of the variable saturation zone formation process and pollutant migration process, clarifying the range of variable saturation porous media and pollutant plume change areas; enhancing the effective information of the images of the variable saturation zone formation and pollutant migration process, magnifying image details in near-boundary areas to improve brightness distortion; and using the information gradient differences of continuous images to characterize the structural features and interface composition of the variable saturation zone formation process and the behavior of the pollutant migration process, obtaining an initial image. This invention has the advantages of high efficiency, practicality, simplicity, and accuracy. Attached image description:
[0037] Figure 1 This is a schematic diagram of the process of the present invention.
[0038] Figure 2 This is a schematic diagram of the image interpretation system of the present invention. Detailed implementation method:
[0039] To clearly illustrate the technical features of this solution, the invention will be described in detail below through specific implementation methods and in conjunction with the accompanying drawings.
[0040] like Figure 1-2 The image interpretation method shown herein, used for identifying contaminant migration in variable-saturation porous media, includes the following steps:
[0041] S1 continuously acquires images of the formation process of the variable saturation zone and the migration process of pollutants, and performs color space conversion and grayscale processing;
[0042] S2, based on grayscale spatial information elements and color spatial information elements, performs image preprocessing, and adjusts the overall changes and local fine features of the variable saturation band formation process and pollutant migration process to clarify the range of variable saturation porous media and pollutant plume change areas;
[0043] S3 enhances effective image information on the formation of variable saturation bands and pollutant migration processes, and magnifies image details in near-boundary areas to improve brightness distortion;
[0044] S4. By utilizing the information gradient differences of continuous images, the structural features and interface composition of the variable saturation zone formation process and the behavior of pollutant migration process are characterized to obtain the initial image.
[0045] S5 performs a difference operation on the acquired process image and the initial image, converts the pixel values into real values, and combines the water saturation calculation formula and the pollution saturation calculation formula to quantitatively evaluate the representative elements of the variable saturation porous media and its pollution migration, accurately quantitatively identify the pollution in the variable saturation porous media, and accurately identify the migration behavior of pollutants in the variable saturation porous media.
[0046] The process of continuously acquiring images of the formation of the variable saturation zone and the migration of pollutants, followed by color space conversion and grayscale processing, includes the following steps:
[0047] S1.1, Adjust camera settings, exposure time and frame rate to change the camera's dynamic range;
[0048] S1.2, automatically capture continuous images of the variable saturation band formation process and pollutant migration process, and set and store the pixel format of the images in MONO8 and RGB formats.
[0049] Image preprocessing based on grayscale and color spatial information elements, adjusting for overall changes and local fine features in the formation process of the variable saturation band and the pollutant migration process, and clarifying the range of the variable saturation porous medium and the pollution plume variation area includes the following steps:
[0050] S2.1 performs morphological processing on the grayscale image, expanding or shrinking white pixel blocks, connecting adjacent white areas, and eliminating isolated white areas; adjusting the shape of the morphological structural unit to a rectangle, ellipse, or cross shape, adjusting the kernel width and height, and increasing the number of iterations, thereby extracting image components that are highly correlated with variable saturation porous media and contaminated areas.
[0051] S2.2, Set the corresponding thresholds to binarize the grayscale images of the formation of variable saturation bands and the pollution process at multiple brightness levels. In order to avoid setting the minimum threshold, the maximum threshold, the filter kernel height and width settings for image smoothness, and the threshold offset settings under the condition of obvious black and white differences, so as to clarify the small pixel differences between the variable saturation band, the pollution plume range boundary and the medium.
[0052] S2.3 Noise suppression is performed while preserving the inherent details of the variable saturation band and pollutant migration characteristics. Multiple filter types are used to adjust the height, width and number of filter kernels for the image, and the smoothing value in the corresponding direction is increased to improve the smoothness of the image. At the same time, the edge clarity range is improved by changing the edge threshold range, thereby improving the effectiveness and reliability of identifying the formation of the variable saturation band and the pollution process.
[0053] The binarization process includes hard thresholding, Gaussian binarization, and mean binarization.
[0054] Enhancing the effective image information of variable saturation band formation and pollutant migration processes, and magnifying image details in near-boundary regions to improve brightness distortion, includes the following steps:
[0055] S3.1, change the sharpening kernel size and sharpening coefficient to determine the local change area and provide sharpening intensity, combined with the adjustment coefficient to control the contrast to change the difference between the variable saturation band, the pollution migration area and the medium in the image;
[0056] S3.2, set gain and compensation to change the overall pixel brightness of the image to achieve brightness correction, and adjust the Gamma value to increase the proportion of image units in areas of light and dark difference, perform non-linear color editing on the image, increase the contrast between the variable saturation band, the pollution migration area and the medium, thereby magnifying details.
[0057] By utilizing the information gradient differences in consecutive images, the structural features and interface composition of the variable saturation band formation process, as well as the behavior of pollutant migration, are characterized to obtain the initial image. This process includes the following steps:
[0058] S4.1, phase interface identification is performed on the image of the variable saturation zone formation process. A machine learning image segmentation algorithm is introduced to segment the phase interface on the variable saturation zone profile. A limited number of manual annotations are used to train the classifier and automatically segment the remaining interface data. The segmentation results of the gas-solid interface and the liquid-solid interface are optimized to generate the region area and obtain the ratio of liquid phase area to gas phase area in the pores of the variable saturation porous medium.
[0059] S4.2, detect and extract the saturation band and pollution migration process edge points of the grayscale image. After filtering and boundary enhancement processing of the image, the vertex edge is detected and located. Based on the image gradient that is different between the extreme point and the surrounding nearby pixels, the grayscale change of the region is calculated to obtain the local extreme point, thereby adjusting the suppression threshold to determine the vertex edge.
[0060] S4.3 extracts brightness and color from the color image by performing saturation band and pollution migration process, calculates the relationship between image pixel values and true values, uses pixel values to represent the range of the region of interest, and sets the corresponding center point coordinates and outer and inner diameters. Then, through color channel conversion, sets a reasonable threshold to suppress target and background areas, selects RGB / HSL channels to measure the average, maximum and standard deviation of color and brightness, thereby confirming the numerical changes of multiple channels under different color spaces.
[0061] The formula for calculating the water saturation is:
[0062]
[0063] Where, S(x, y) is the water saturation at each pixel, and L s(x，y) L is the brightness value of this pixel. air L represents the brightness value when the pores of a porous medium are completely filled with air. saturated This represents the brightness value when the porous medium is fully saturated with water.
[0064] The formula for calculating pollution saturation is:
[0065]
[0066] Where F(x, y) is the contamination saturation of each pixel, and L f(x，y) L is the brightness value of a pixel. background L represents the background brightness value before any pollutants are introduced. f，0 This represents the maximum brightness value when the pollutants are introduced into the porous medium during the initial stage of the experiment.
[0067] The image interpretation system of the image interpretation method includes:
[0068] An image acquisition device, which is an external device, is placed at the front end of an indoor sand tank or sand column plexiglass plate. It is used to detect the area where the variable saturation zone forms and pollutants migrate. When the variable saturation zone begins to form and pollutants begin to migrate, the image is acquired in real time and stored in a memory.
[0069] A computer medium includes a memory, a processor, and a decoder; the memory stores continuous images acquired by an image acquisition device; the processor processes images of variable saturation band formation and pollutant migration processes; and the decoder calculates representative elements from the images of the variable saturation porous medium and its pollutant migration obtained by the processor.
[0070] The working principle of the image interpretation method for identifying pollutant migration in variable-saturated porous media in this invention embodiment is as follows: the precise characterization of the structural features and interfaces of variable-saturated porous media is aggregated into a quantitative expression, and the pollutants in variable-saturated porous media are accurately quantitatively identified. It can reveal the spatial heterogeneity of water saturation distribution and its phase distribution characteristics in porous media from a microscopic perspective, thereby accurately identifying the migration behavior of pollutants in variable-saturated porous media. Unlike previous destructive sampling and fixed local measurement methods, it can achieve precise identification at the nanometer-level pixel level for any region, and can also achieve rapid response to millions or more data features and results, balancing data processing efficiency and processing accuracy.
[0071] In the overall scheme, the image interpretation method includes the following steps: continuously acquiring images of the variable saturation zone formation process and pollutant migration process, performing color space conversion and grayscale processing; performing image preprocessing based on grayscale and color space information elements, adjusting for the overall changes and local fine features of the variable saturation zone formation process and pollutant migration process, clarifying the range of variable saturation porous media and pollutant plume change areas; enhancing the effective information of the images of variable saturation zone formation and pollutant migration process, magnifying image details in near-boundary areas to improve brightness distortion; utilizing the information gradient differences of continuous images to characterize the structural features and interface composition of the variable saturation zone formation process and the expression of pollutant migration behavior, obtaining the initial image; performing a difference operation between the acquired process images and the initial image, converting pixel values into true values, and quantitatively evaluating the representative elements of variable saturation porous media and its pollutant migration by combining the water saturation calculation formula and the pollutant saturation calculation formula, accurately quantitatively identifying the pollution in variable saturation porous media, and accurately identifying the migration behavior of pollutants in variable saturation porous media.
[0072] By changing camera settings, exposure time, frame rate, etc., the dynamic range of the camera is improved, and continuous images of the formation process of variable saturation bands and the migration process of pollutants are automatically captured. The continuous image pixel format is set and stored in MONO8 and RGB formats, thus providing a basis for image interpretation.
[0073] Preferably, image preprocessing based on grayscale and color space information elements includes the following steps: morphological processing of the grayscale image, expanding or shrinking white pixel blocks, connecting adjacent white areas, and eliminating isolated white areas; adjusting the shape of the morphological structural units to rectangles, ellipses, or crosses, adjusting the kernel width and height, and increasing the number of iterations to extract image components highly correlated with variable saturation porous media and contamination areas; setting appropriate thresholds to binarize grayscale images of variable saturation band formation and contamination processes at multiple brightness levels, avoiding minimum threshold settings, maximum threshold settings, filter kernel height and width settings for image smoothness, and threshold offset settings under conditions of significant black-and-white differences, thereby clarifying the small pixel differences between the boundary of variable saturation bands and contamination plumes and the medium, reflecting overall changes and local small features to a certain extent; noise suppression while preserving the inherent details of variable saturation bands and contaminant migration characteristics, using multiple filter types to adjust the filter kernel height, width, and number settings, increasing the smoothing value in the corresponding direction, and improving the smoothness of the image; simultaneously improving the edge clarity range by changing the edge threshold range, thereby improving the effectiveness and reliability of identifying variable saturation band formation and contamination processes.
[0074] Binarization processing mainly includes hard thresholding, Gaussian binarization, and mean binarization, which can further improve the accuracy of the processing.
[0075] Preferably, enhancing the effective image information of the formation of variable saturation bands and the pollutant migration process, and magnifying image details in near-boundary areas to improve brightness distortion includes the following steps: changing the sharpening kernel size and sharpening coefficient to determine the local change area and provide sharpening intensity; combining this with the contrast control adjustment coefficient to change the differences between the variable saturation band, the pollutant migration area, and the medium in the image; simultaneously setting gain and compensation to change the overall pixel brightness of the image to achieve brightness correction; adjusting the Gamma value to increase the proportion of image units in areas of light and dark difference; performing non-linear color editing on the image to increase the contrast between the variable saturation band, the pollutant migration area, and the medium, thereby magnifying details.
[0076] In the process of phase interface identification in the image of the variable saturation zone formation process, due to the complex interlacing pore channels of porous media, the operation steps to determine the phase distribution ratio in the pores from a microscopic perspective are cumbersome. From a macroscopic perspective, the soil particle wetting degree of liquid-filled pores and gas-filled pores is different, and their interfaces should show obvious inner boundary edges. However, the interface edges are irregular, making it difficult to directly measure the bubble area. Therefore, a machine learning image segmentation algorithm is introduced to segment the phase interfaces on the variable saturation zone profile. A limited number of manual annotations are used to train a classifier and automatically segment the remaining interface data, optimize the segmentation results of the gas-solid interface and the liquid-solid interface, generate the region area, and obtain the proportion of liquid phase area and gas phase area in the pores of the variable saturation porous media.
[0077] For grayscale images, edge detection and extraction of variable saturation bands and contamination migration processes are performed. Since the optical properties of porous media differ before and after immersion in water and contamination, there are obvious changes in image pixel intensity under reflection / transmission imaging. This information gradient difference can be used for accurate calculation. After filtering and boundary enhancement processing of the image, vertex edges are detected and located. Based on the different image gradients between extreme points and surrounding pixels, the grayscale changes in the region are calculated to obtain local extreme points. The threshold is adjusted and suppressed to determine the vertex edges.
[0078] For color images, saturation bands, brightness and color extraction during contamination migration are performed. The relationship between image pixel values and true values is calculated. Pixel values are used to represent the range of the region of interest. Parameters such as center point coordinates, outer diameter, and inner diameter are set. Then, through color channel conversion, reasonable thresholds are set to suppress target and background areas. The average, maximum, and standard deviation of color and brightness are measured using RGB / HSL channels to confirm the numerical changes of multiple channels under different color spaces.
[0079] Furthermore, images taken before the variable saturation band begins to form and before pollutants begin to migrate are selected as background values, and differential operations are performed with continuous images of the variable saturation band formation and pollutant migration process. In this application, HSL color space conversion is performed, and water saturation and pollution saturation are selected as representative elements. The values of each pixel block in the entire domain are calculated by extracting the L channel. The calculation formulas for different pollutants are not exactly the same. In this application, the pollutants are fluorescent Escherichia coli and fluorescent polystyrene nanoplastic microspheres.
[0080] Preferably, the formula for calculating the water saturation in this application is as follows:
[0081]
[0082] Where, S(x, y) is the water saturation at each pixel, and L s(x，y) L is the brightness value of this pixel. air L represents the brightness value when the pores of a porous medium are completely filled with air. saturated This represents the brightness value when the porous medium is fully saturated with water.
[0083] The formula for calculating pollution saturation is:
[0084]
[0085] Where F(x, y) is the contamination saturation of each pixel, and L f(x，y) L is the brightness value of a pixel. background L represents the background brightness value before any pollutants are introduced. f，0 The value represents the maximum brightness when the pollutant is introduced into the porous medium at the initial stage of the experiment; the maximum brightness occurs when the fluorescent pollutant reaches saturation in the porous medium.
[0086] The image interpretation system for the image interpretation method includes: an image acquisition device, a memory, a processor, and an interpreter; the image acquisition device is an external device placed at the front end of the indoor sand tank / sand column plexiglass plate, used to detect the area where the variable saturation zone forms and pollutants migrate; when the variable saturation zone begins to form and pollutants begin to migrate, the image is acquired in real time and the acquired image is stored in the memory.
[0087] The memory, processor, and interpreter are all installed in the computer medium. The memory stores continuous images acquired by the image acquisition device, as well as related computer programs and code. This allows the processor to process images of variable saturation band formation and pollutant migration processes using the relevant calculation programs and code stored in the memory. The interpreter can perform representative element calculations on the variable saturation porous medium and its pollutant migration images obtained by the processor using the relevant calculation programs and code stored in the memory.
[0088] In summary, the image interpretation method for identifying pollutant migration in variable-saturated porous media in this invention precisely characterizes the structural features and interfaces of variable-saturated porous media into a quantitative expression, enabling accurate quantitative identification of pollutants in these media. It can reveal the spatial heterogeneity of water saturation distribution and its phase distribution characteristics from a microscopic perspective, thereby accurately identifying the migration behavior of pollutants in variable-saturated porous media. Unlike previous destructive sampling and fixed local measurement methods, this method can achieve precise identification at the nanometer pixel level for any region and can also achieve rapid response to millions or more data features and results, balancing data processing efficiency and accuracy.
[0089] The above specific embodiments should not be construed as limiting the scope of protection of the present invention. For those skilled in the art, any alternative improvements or modifications made to the embodiments of the present invention shall fall within the scope of protection of the present invention.
[0090] Any aspects of this invention not described in detail are well-known to those skilled in the art.
Claims
1. An image interpretation method for identifying pollutant migration in variable-saturation porous media, characterized in that, The image interpretation method includes the following steps: S1 continuously acquires images of the formation process of the variable saturation zone and the migration process of pollutants, and performs color space conversion and grayscale processing; S2, based on grayscale spatial information elements and color spatial information elements, performs image preprocessing, and adjusts the overall changes and local fine features of the variable saturation band formation process and pollutant migration process to clarify the range of variable saturation porous media and pollutant plume change areas; S3 enhances effective image information on the formation of variable saturation bands and pollutant migration processes, and magnifies image details in near-boundary areas to improve brightness distortion; S4. By utilizing the information gradient differences of continuous images, the structural features and interface composition of the variable saturation zone formation process and the behavior of pollutant migration process are characterized to obtain the initial image. S5 performs a difference operation on the acquired process image and the initial image, converts the pixel value into the real value, and combines the water saturation calculation formula and the pollution saturation calculation formula to quantitatively evaluate the representative elements of the variable saturation porous media and its pollution migration, accurately quantitatively identify the pollution in the variable saturation porous media, and accurately identify the migration behavior of pollutants in the variable saturation porous media. By utilizing the information gradient differences in consecutive images, the structural features and interface composition of the variable saturation band formation process, as well as the behavior of pollutant migration, are characterized to obtain the initial image. This process includes the following steps: S4.1, phase interface identification is performed on the image of the variable saturation zone formation process. A machine learning image segmentation algorithm is introduced to segment the phase interface on the variable saturation zone profile. A limited number of manual annotations are used to train the classifier and automatically segment the remaining interface data. The segmentation results of the gas-solid interface and the liquid-solid interface are optimized to generate the region area and obtain the ratio of liquid phase area to gas phase area in the pores of the variable saturation porous medium. S4.2, detect and extract the saturation band and pollution migration process edge points of the grayscale image. After filtering and boundary enhancement processing of the image, the vertex edge is detected and located. Based on the image gradient that is different between the extreme point and the surrounding nearby pixels, the grayscale change of the region is calculated to obtain the local extreme point, thereby adjusting the suppression threshold to determine the vertex edge. S4.3 extracts brightness and color from the color image by performing saturation band and pollution migration process, calculates the relationship between image pixel values and true values, uses pixel values to represent the range of the region of interest, and sets the corresponding center point coordinates and outer and inner diameters. Then, through color channel conversion, a reasonable threshold is set to suppress the target and background areas. The RGB / HSL channels are selected to measure the average, maximum and standard deviation of color and brightness, thereby confirming the numerical changes of multiple channels under different color spaces.
2. The image interpretation method for identifying pollutant migration in variable-saturation porous media according to claim 1, characterized in that, The process of continuously acquiring images of the formation of the variable saturation zone and the migration of pollutants, followed by color space conversion and grayscale processing, includes the following steps: S1.1, Adjust camera settings, exposure time and frame rate to change the camera's dynamic range; S1.2, automatically capture continuous images of the variable saturation band formation process and pollutant migration process, and set and store the pixel format of the images in MONO8 and RGB formats.
3. The image interpretation method for identifying pollutant migration in variable-saturation porous media according to claim 1, characterized in that, Image preprocessing based on grayscale and color spatial information elements, adjusting for overall changes and local fine features in the formation process of the variable saturation band and the pollutant migration process, and clarifying the range of the variable saturation porous medium and the pollution plume variation area includes the following steps: S2.1 performs morphological processing on the grayscale image, expanding or shrinking white pixel blocks, connecting adjacent white areas, and eliminating isolated white areas; adjusting the shape of the morphological structural unit to a rectangle, ellipse, or cross shape, adjusting the kernel width and height, and increasing the number of iterations, thereby extracting image components that are highly correlated with variable saturation porous media and contaminated areas. S2.2, Set the corresponding thresholds to binarize the grayscale images of the formation of variable saturation bands and the pollution process at multiple brightness levels. In order to avoid setting the minimum threshold, the maximum threshold, the filter kernel height and width settings for image smoothness, and the threshold offset settings under the condition of obvious black and white differences, so as to clarify the small pixel differences between the variable saturation band, the pollution plume range boundary and the medium. S2.3 Noise suppression is performed while preserving the inherent details of the variable saturation band and pollutant migration characteristics. Multiple filter types are used to adjust the height, width and number of filter kernels for the image, and the smoothing value in the corresponding direction is increased to improve the smoothness of the image. At the same time, the edge clarity range is improved by changing the edge threshold range, thereby improving the effectiveness and reliability of identifying the formation of the variable saturation band and the pollution process.
4. The image interpretation method for identifying pollutant migration in variable-saturation porous media according to claim 3, characterized in that: The binarization process includes hard thresholding, Gaussian binarization, and mean binarization.
5. The image interpretation method for identifying contaminant migration in variable-saturation porous media according to claim 1, characterized in that, Enhancing the effective image information of variable saturation band formation and pollutant migration processes, and magnifying image details in near-boundary regions to improve brightness distortion, includes the following steps: S3.1, change the sharpening kernel size and sharpening coefficient to determine the local change area and provide sharpening intensity, combined with the adjustment coefficient to control the contrast to change the difference between the variable saturation band, the pollution migration area and the medium in the image; S3.2, set gain and compensation to change the overall pixel brightness of the image to achieve brightness correction, and adjust the Gamma value to increase the proportion of image units in areas of light and dark difference, perform non-linear color editing on the image, increase the contrast between the variable saturation band, the pollution migration area and the medium, thereby magnifying details.
6. The image interpretation method for identifying contaminant migration in variable-saturation porous media according to claim 1, characterized in that, The formula for calculating the water saturation is: ； in, The water saturation at each pixel. This is the brightness value of that pixel. This represents the brightness value when the pores of the porous medium are completely filled with air. This represents the brightness value when the porous medium is fully saturated with water.
7. The image interpretation method for identifying contaminant migration in variable-saturation porous media according to claim 1, characterized in that, The formula for calculating pollution saturation is: ； in, The contamination saturation of each pixel, The brightness value of a pixel. The brightness background value is the value before any pollutants are introduced. This represents the maximum brightness value when the pollutants are introduced into the porous medium during the initial stage of the experiment.
8. The image interpretation method for identifying contaminant migration in variable-saturation porous media according to claim 1, characterized in that, The image interpretation system of the image interpretation method includes: An image acquisition device, which is an external device, is placed at the front end of an indoor sand tank or sand column plexiglass plate. It is used to detect the area where the variable saturation zone forms and pollutants migrate. When the variable saturation zone begins to form and pollutants begin to migrate, the image is acquired in real time and stored in a memory. A computer medium includes a memory, a processor, and a decoder; the memory stores continuous images acquired by an image acquisition device; the processor processes images of variable saturation band formation and pollutant migration processes; and the decoder calculates representative elements from the images of the variable saturation porous medium and its pollutant migration obtained by the processor.