Method for calculating relative spatial distribution of particulate organic matter and pore structure in ct images

Abstract

The application relates to the technical field of soil science, in particular to a method for calculating the relative spatial distribution of particle organic matter and pore structure in a CT image, which comprises the following steps: S1: CT scanning of a soil sample and image preprocessing to obtain a preprocessed CT image; S2: based on machine learning, particle organic matter and pore structure are identified and segmented from the preprocessed CT image; S3: morphological characteristics of the pore structure are calculated; S4: morphological characteristics of the particle organic matter are calculated; S5: the relative spatial distribution of the particle organic matter and the pore structure is analyzed, including calculation of the distance distribution of the particle organic matter to the surface of the pore structure. The application not only calculates the morphological parameters of the pore structure and the organic matter, but also calculates the distribution of the particle organic matter relative to the pore structure and the content of the particle organic matter connected with the pores, and can directly represent the interaction mechanism between the two.

Description

Technical Field

[0001] This invention relates to the field of soil science and technology, specifically to a method for calculating the relative spatial distribution of particulate organic matter and pore structure in CT images. Background Technology

[0002] In farmland soils, soil particulate organic matter (POM) is mainly composed of the decomposition products of straw, roots, and leaves, and is the most active component of soil organic matter. Increasing the content of particulate organic matter is of great significance for enhancing carbon sequestration in farmland soils. Currently, CT technology can accurately quantify the morphological characteristics and spatial distribution of particulate organic matter. Existing technology publication number CN117218437B provides a method for in-situ quantitative quantification of soil particulate organic matter by combining CT technology with machine learning, providing a powerful tool for accurately quantifying the spatial distribution of POM in soil.

[0003] Soil particulate organic matter is mainly distributed in pores and is physically protected by aggregates. Although a combined particle size and density grouping method can be used to separate particulate organic matter between and within the pores of aggregates, the in-situ spatial relative distribution of particulate organic matter and pore structure is still not quantitatively determined. Summary of the Invention

[0004] To address the problems in the prior art, this invention provides a method for calculating the relative spatial distribution of particulate organic matter and pore structure in CT images.

[0005] The technical solution adopted by this invention to solve its technical problem is: a method for calculating the relative spatial distribution of particulate organic matter and pore structure in CT images, comprising the following steps:

[0006] S1: Perform CT scans on soil samples and preprocess the images to obtain preprocessed CT images;

[0007] S2: Based on machine learning, it identifies and segments particulate organic matter and pore structure from preprocessed CT images;

[0008] S3: Calculate the morphological characteristics of the pore structure;

[0009] S4: Calculate the morphological characteristics of particulate organic matter;

[0010] S5: Analyze the relative spatial distribution of particulate organic matter and pore structure, including calculating the distance distribution from particulate organic matter to the surface of pore structure;

[0011] S6: Analyze and extract particulate organic matter connected to the pore structure.

[0012] Preferably, S2 further includes:

[0013] Supervised machine learning tools were used to classify each pixel in the preprocessed CT images, with classification labels including at least soil matrix, particulate organic matter, and pores.

[0014] Based on the classification results, binary images corresponding to particulate organic matter and pore structure are generated respectively.

[0015] Preferably, in S2, the particle organic matter image and pore structure image obtained after classification are further post-processed, including: median filtering, gray value inversion, local threshold binarization, and particle organic matter sieving based on volume threshold.

[0016] Preferably, in S3, the morphological characteristics of the pore structure are calculated, including the volume, surface area, average width, sphericity, Euler number, and pore size distribution of one or more of the following: total pores, connected pores, isolated pores, top-connected pores, and permeable pores.

[0017] Connected pores and isolated pores can be classified in any of the following ways:

[0018] (i) Define the largest pore as a connected pore, and the rest as isolated pores.

[0019] (ii) Define pores that are connected to the boundary of the region of interest as connected pores, and define pores that are completely located inside the region of interest as isolated pores.

[0020] Preferably, in S4, the morphological characteristics of particulate organic matter are calculated, including the volume, surface area, average width, sphericity, and particle size distribution of the particulate organic matter.

[0021] Preferably, the analysis of the relative spatial distribution of particulate organic matter and pore structure in S5 further includes:

[0022] Generate a three-dimensional distance distribution map representing the distances from the soil matrix and particulate organic matter to the nearest pore surface;

[0023] Multiplying the three-dimensional distance distribution map with the normalized particulate organic matter image yields a distance distribution map of particulate organic matter relative to the pore surface, which is used to characterize the degree to which particulate organic matter is physically protected by pores.

[0024] Preferably, S5 also includes:

[0025] By interchangeding the granular organic matter and pore structure, a three-dimensional distance distribution map representing the soil matrix and pores to the nearest granular organic matter surface is generated. This distance distribution map is then multiplied with the normalized pore image to obtain the distance distribution map of pores relative to the granular organic matter surface.

[0026] Preferably, in step S6, the analysis and extraction of particulate organic matter connected to the pore structure specifically includes:

[0027] Dilate the pore structure image by one voxel;

[0028] Mathematical operations were performed on the expanded pore structure image and the particulate organic matter image to obtain the intersection of the two;

[0029] Using the intersecting portion as the labeled image and the original particulate organic matter image as the mask image, three-dimensional morphological reconstruction is performed to extract particulate organic matter that is fully connected to the pore structure.

[0030] Preferably, S6 also includes:

[0031] The total pore image was replaced with connected pore images and isolated pore images, respectively, and the particulate organic matter connected to connected pores and the particulate organic matter connected to isolated pores were extracted.

[0032] When the same particle of organic matter is connected to both connected pores and isolated pores, it is classified as being connected to connected pores.

[0033] The particulate organic matter distributed within the soil matrix is ​​obtained by subtracting the particulate organic matter connected to interconnected pores and the particulate organic matter connected to isolated pores from the total particulate organic matter.

[0034] Preferably, the method further includes S7: classifying particulate organic matter into heat-sensitive POM and heat-stable POM based on temperature response characteristics, specifically including the following steps:

[0035] S71: During CT scanning, a controlled heating procedure is implemented on the soil sample. Specifically, the soil sample is placed in a CT scanning device equipped with an in-situ heating stage, and the temperature is gradually increased from room temperature to a predetermined maximum temperature (e.g., 200℃) at a constant heating rate. During the heating process, CT scans are performed at eight temperature nodes: 25℃, 50℃, 75℃, 100℃, 125℃, 150℃, 175℃, and 200℃, acquiring CT image sequences of the same soil sample at different temperatures. At each temperature node, the sample is held at the temperature for 5-10 minutes to allow the internal temperature to reach uniform equilibrium before scanning. Scanning parameters are kept consistent (voltage 100kV, current 100μA) to ensure the comparability of grayscale values ​​in CT images at different temperature nodes.

[0036] S72: Based on machine learning, the particulate organic matter in the CT images at each temperature node is identified and segmented. The identification and segmentation method is the same as that in S2, including using supervised machine learning tools (such as ilastik software) for pixel classification, and generating binarized images of particulate organic matter at each temperature node through post-processing such as median filtering and local thresholding.

[0037] S73: Track the grayscale value changes of the same particulate organic matter fragment at different temperature nodes and establish a grayscale value-temperature response curve. Specifically, the operation is as follows: First, using the particulate organic matter image at 25℃ as a reference, the CT images at each temperature node are spatially aligned with the reference image using 3D image registration technology to correct for minor displacements that may be caused by thermal expansion. Then, for each voxel or particulate fragment identified as particulate organic matter, its average grayscale value at each temperature node is extracted, and a grayscale value-temperature response curve is plotted.

[0038] S74: Calculate the grayscale value attenuation characteristic parameters; Due to the difference in thermal stability between fresh POM (undecomposed or semi-decomposed plant residues, mainly composed of thermally unstable components such as cellulose and hemicellulose) and humified POM (stable organic matter formed after long-term microbial decomposition, rich in thermally stable components such as lignin and aromatic compounds), both undergo different degrees of thermal decomposition and density changes during heating, which are reflected in the amplitude and rate of change of CT grayscale values; Specifically, cellulose in fresh POM undergoes thermal decomposition in the range of 200-350℃, while lignin and aromatic compounds in humified POM undergo thermal decomposition in the range of 200-350℃. Significant decomposition only occurs within the 400-650℃ range. Therefore, during controlled heating in the medium-low temperature range (25-200℃), the rate and magnitude of gray value decrease in fresh POM are significantly greater than those in humified POM. Based on this principle, the gray value decay characteristic parameters of each organic matter fragment are calculated, including: initial gray value (average gray value at 25℃), final gray value (average gray value at 200℃), total gray value decay rate (calculated as (initial gray value - final gray value) / initial gray value × 100%), and the gray value decay half-temperature point (the temperature corresponding to when the gray value drops to half of the initial value).

[0039] S75: Based on the grayscale attenuation characteristic parameter, particulate organic matter is classified into heat-sensitive POM (corresponding to fresh POM) and heat-stable POM (corresponding to humified old POM). The specific criteria are: particulate organic matter fragments with a total grayscale attenuation rate greater than 20% are classified as heat-sensitive POM; particulate organic matter fragments with a total grayscale attenuation rate less than or equal to 20% are classified as heat-stable POM. The determination of this threshold is based on the following: POM samples from known sources (such as artificially added fresh corn stalks)... The controlled-heat CT scan experiment was conducted on the debris labeled as fresh POM and the humified POM that had been cultured indoors for 18 months labeled as old POM. The average total attenuation rate of gray value of fresh POM (n=150 fragments) was 32.6%±5.8%, and the average total attenuation rate of gray value of old POM (n=150 fragments) was 10.3%±3.2%, with a significant difference between the two (p<0.01). Based on this, the median value of 20% was taken as the discrimination threshold, and the classification accuracy reached 91.3%.

[0040] Preferably, S7 further includes S76: calculating the spatial distribution characteristics of heat-sensitive POM and heat-stable POM and their relative spatial relationship with the pore structure, respectively; specifically including: calculating the morphological characteristics of the two types of POM, such as volume, surface area, average width, sphericity, and particle size distribution; generating distance distribution maps from heat-sensitive POM to the pore surface and from heat-stable POM to the pore surface according to the method in S5, and analyzing the difference in the degree of physical protection of the two types of POM by the pores; and extracting heat-sensitive POM connected to connected pores and heat-stable POM connected to isolated pores according to the method in S6, and analyzing the difference in the occurrence position of the two types of POM in the pore system.

[0041] Preferably, the method is applied to CT image analysis of soil samples at the soil column scale and / or aggregate scale; and particulate organic matter can be replaced with plant roots, and pore structure can be replaced with biological pores to analyze the root-pore interaction mechanism.

[0042] The beneficial effects of this invention are:

[0043] (1) This invention not only calculates the morphological parameters of pore structure and organic matter, but also calculates the distribution of particulate organic matter relative to pore structure and the content of particulate organic matter connected to pores, which can intuitively characterize the interaction mechanism between the two.

[0044] (2) By introducing a controlled heating program and temperature response characteristic analysis, this invention can distinguish particulate organic matter into heat-sensitive POM and heat-stable POM from the perspective of chemical composition and thermal stability. This enables in-situ, non-destructive evaluation of the degree of POM decomposition and stability, overcoming the limitation of traditional morphological classification methods that make it difficult to accurately distinguish POM with a high degree of humification. Attached Figure Description

[0045] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0046] Figure 1 This is a flowchart illustrating the segmentation of porous and particulate organic matter in Embodiment 1 of the present invention.

[0047] Figure 2 This illustrates the three-dimensional distribution of isolated pores, interconnected pores, top-connected pores, and permeable pores in Embodiment 1 of the present invention.

[0048] Figure 3 This is a particle size distribution diagram of the particulate organic matter in Example 1 of the present invention;

[0049] Figure 4 This is a flowchart illustrating the calculation of the distance distribution from particulate organic matter (POM) to pores and the distance distribution from pores to POM in Embodiment 1 of the present invention.

[0050] Figure 5 This is a diagram showing the distance distribution from particulate organic matter to pores in Example 1 of the present invention;

[0051] Figure 6 This is a diagram showing the distance distribution from pores to particulate organic matter in Example 1 of the present invention;

[0052] Figure 7 This is a two-dimensional diagram showing the connection between particulate organic matter and interconnected and isolated pores, as well as their distribution in the soil matrix, in Example 1 of the present invention.

[0053] Figure 8 This is a three-dimensional distribution diagram of particulate organic matter in the soil matrix, showing its connection to interconnected and isolated pores, as well as its distribution within the soil matrix, in Example 1 of the present invention.

[0054] Figure 9 This is a flowchart illustrating the segmentation of porous and particulate organic matter in Embodiment 2 of the present invention;

[0055] Figure 10 This is a flowchart illustrating the separation of connected pores and isolated pores in Embodiment 2 of the present invention;

[0056] Figure 11 This is a three-dimensional distribution diagram of isolated pores and connected pores in Embodiment 2 of the present invention;

[0057] Figure 12 This is a particle size distribution diagram of the particulate organic matter in Example 2 of the present invention;

[0058] Figure 13 This is a flowchart illustrating the calculation of the distance distribution from particulate organic matter (POM) to pores and the distance distribution from pores to POM in Embodiment 2 of the present invention.

[0059] Figure 14 This is a diagram showing the distance distribution from particulate organic matter to pores in Example 2 of the present invention;

[0060] Figure 15 This is a diagram showing the distance distribution from pores to particulate organic matter in Example 2 of the present invention;

[0061] Figure 16 This is a two-dimensional diagram showing the connection between particulate organic matter and interconnected and isolated pores, as well as their distribution in the soil matrix, in Example 2 of the present invention.

[0062] Figure 17 This is a three-dimensional distribution diagram of particulate organic matter in the soil matrix, showing its connection to interconnected and isolated pores, as well as its distribution within the soil matrix, in Example 2 of the present invention.

[0063] Figure 18 This is a flowchart of the present invention. Detailed Implementation

[0064] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0065] The following examples all involve randomly collecting top 0-10cm soil samples from the Longkang Farm tillage and fertilization experimental base in Huaiyuan County, Anhui Province. The sampling methods for Examples 1 and 2 differ; in Example 1, the sample is a soil column, taken from a 100cm soil column. 3 PVC ring cutter sampling: The sample in Example 2 was an agglomerate, which was selected from the experimental site with a diameter of 3-5 mm.

[0066] Example 1:

[0067] This embodiment uses 100cm [text missing] from the tillage and fertilization experimental base of Longkang Farm, Huaiyuan County, Anhui Province. 3 Soil samples from the top 0-10cm layer were randomly collected using a PVC ring cutter and then subjected to CT scanning.

[0068] S1: CT scans were performed, and the images were reconstructed using Datos|x2Rec software, converting them into an analyzable TIFF format. The soil column sample image resolution was 30 μm. The reconstructed TIFF images were imported into ImageJ / Fiji software. Adjust-Brightness / Contrast was used to adjust the image brightness and contrast, and a median filter with a radius of 2 voxels was used for noise reduction. Then, an anti-sharpening mask filter was used to reduce local volumetric effects. The central region of the image was cropped as the region of interest (ROI), with the ROI size being a cylinder with a diameter of 1350 voxels and a height of 1300 voxels. The image was saved as a TIFF sequence.

[0069] S2: Open the pixel classification workflow in the Ilastik software, select Add a Single 3D / 4D Volume from Sequence, import the image from S1, and select pixel features including grayscale value, edge information, and texture features, with Gaussian smoothing standard deviations of δ=0.3, 0.7, and 1.0, respectively. Classify the original image into soil matrix, pores, and particulate organic matter, and add three labels accordingly, each corresponding to a pixel category. Then, use the pen tool to add annotations to each pixel category and perform classification training (LiveUpdate). If misclassification occurs, re-label the pixels using the pen or eraser function until the training result is accurate. The exported image data type is Integer8bit, and the image format is TiffSequence.

[0070] The segmented image from Ilastik was imported into Fiji software. Color-SplitChannels was used to split the RGB channels of the image into soil matrix, pores, and granular organic matter. Image sequences containing granular organic matter and pores were retained and filtered using a median filter with a radius of 2 pixels. The image grayscale values ​​were then inverted. The image was binarized using the Sauvola algorithm in AutoLocalThreshold, with default parameter settings. After binarization, the black areas in the image were larger than one voxel (30×30×30μm). 3 The particulate organic matter or pores are sieved. The particulate organic matter is sieved to obtain a sieve volume greater than 50×50×50μm. 3 For the particulate organic matter, the ConnectedComponentsLabeling function of the MorphoLibJ plugin is used to add labels to black pixels. Then, LabelSizeFiltering is used to screen out particulate organic matter with a volume greater than 5 voxels and convert it into a binary image. Figure 1 Images of pores and granular organic matter obtained after image segmentation of grayscale images are listed.

[0071] S3: Calculate the basic characteristics of the pores using the AnalyzeRegions3D function of the MorphoLibJ plugin: volume, surface area, average width, sphericity, and Euler number; as shown in Table 1.

[0072] Table 1. Pore structure characteristic parameters at the soil column scale

[0073]

[0074] Analyzing the top connecting pores: In Fiji software, the pore image is sequentially reversed (Image-Stacks-Tools-Reverse) and then saved as a TIFF image. Open the SoilJ plugin's ImageAnalyses-PoreSpaceAnalyzer, select the cylindrical ROI (Cylindrical ROI) in the pop-up dialog box, and click OK; in the next dialog box, click OK without making any changes; in the third dialog box, select VolCon2Top, leaving other parameters unselected, and click OK; in the fourth dialog box, select the TIFF file to save, and the program will automatically calculate the top connecting pores. The results are saved in the same folder as the TIFF image, automatically generating a folder named "255_Pores," with the top connecting pores saved in a subfolder named "VolumeConnected2Top." Import the top connecting pores into Fiji software, first adjust the image sequence (Image-Stacks-Tools-Reverse), and then perform Process-Binary-MakeBinary processing on the image to make the black pixels in the image represent pores. The characteristics of the top connecting pores are calculated, as shown in Table 1. This method can be used to analyze the organic matter in the top-connected particles.

[0075] Separating Connected and Isolated Pores: The pore with the largest volume is defined as a connected pore, and the rest are considered isolated pores. Connected pores are extracted using the "Purify" function, and then the connected pore image is subtracted from the total pore image (Process-ImageCalculate-Subtract) to obtain the isolated pores. Features of connected and isolated pores are calculated, such as... Figure 1 As shown.

[0076] Analyzing permeability: Save the pore image directly as a TIFF image. Open the ImageAnalyses-PoreSpaceAnalyzer in the SoilJ plugin. In the pop-up dialog box, select the cylindrical ROI (Cylindrical ROI) and click OK. In the next dialog box, click OK without making any changes. In the third dialog box, select CriticalDiameter and PercolatingClusters, leaving other parameters unselected, and click OK. In the fourth dialog box, select the saved TIFF file. The program will automatically calculate the permeability and critical pore diameter. The results are saved in the same folder as the TIFF image, automatically generating a folder named "255_Pores". The permeability pores are saved in a subfolder named "PercolatingVolume", and the critical pore diameter is saved in the ROI file in the subfolder named "Stats". Import the permeability pores into Fiji software and perform Process-Binary-MakeBinary processing on the image to make the black pixels in the image represent pores. Calculate the permeability pore characteristics, as shown in Table 1. Open the ROI file with Notepad. The critical diameter is 6.56 voxels, and the actual size is 197 μm.

[0077] S4: The basic morphological characteristics of the POM (PoM) were calculated using the MorphoLibJ plugin's AnalyzeRegions3D function: volume, surface area, average width, and sphericity, as shown in Table 2. The particle size distribution and average diameter of the POM were calculated using Bonej's Thickness function. The particle size distribution is shown in Table 2. Figure 4 As shown, the average diameter of POM is 118 μm.

[0078] Table 2. Morphological characteristics of particulate organic matter at the soil column scale.

[0079]

[0080] Note: POM connected, POM isolated, and POM matrix refer to particulate organic matter connected to connected pores, connected to isolated pores, and distributed in the soil matrix, respectively.

[0081] S5: (1) Calculate the distance distribution from particulate organic matter to the pore surface: First, the grayscale image is thresholded (0-254). Since the image contains high-density materials such as sand, ginger, and gravel, the grayscale value of this area in the image is high (255). The threshold needs to be processed, i.e., Process-Binary-FillHoles, so that the ROI area is black (grayscale value is 255) to obtain the mask image. The pore image is subtracted from the mask image, and the Euclidean distance transformation (Plugins-Process-ExactEuclideanDistanceTransform (3D)) is performed on the result image to obtain the 3D distance distribution map. This map represents the distance distribution from the scene voxels (including soil matrix and particulate organic matter) to the nearest pore surface. The larger the grayscale value, the farther the distance. The grayscale histogram analysis (Analyze-Histogram) of this distance distribution map is performed to obtain the distance distribution. The average pore distance can be calculated, i.e., the weighted average of the distance from each non-pore voxel to the nearest pore voxel, which is 476μm. Generally, the higher the pore connectivity, the smaller the average pore distance. The grayscale values ​​of the particulate organic matter image are divided by 255 (Process-Math-Divide) to make all grayscale values ​​of the particulate organic matter equal to 1. The normalized particulate organic matter image is then multiplied with the 3D distance distribution map to obtain the distance distribution map of the particulate organic matter relative to the pore surface. The process is as follows: Figure 4 As shown in the figure. A grayscale histogram analysis was performed on this figure to obtain the distance distribution from the particulate organic matter to the pores, as shown below. Figure 5 As shown.

[0082] (2) Calculate the distance distribution from pores to the surface of particulate organic matter: Subtract the particulate organic matter image from the mask image, perform Euclidean distance transformation (Plugins-Process-ExactEuclideanDistanceTransform (3D)) on the resulting image to obtain a 3D distance distribution map. Perform grayscale histogram analysis on this distance distribution map to obtain the distance distribution. The average distance of particulate organic matter can be calculated, which is the weighted average distance from each non-porous voxel to the nearest pore voxel, which is 515 μm. Generally speaking, the higher the content of particulate organic matter, the smaller the average distance of particulate organic matter. Divide the grayscale value of the pore image by 255 (Process-Math-Divide) to make all pore grayscale values ​​1. Multiply the normalized pore image with the 3D distance distribution map to obtain the distance distribution map of pores relative to the surface of particulate organic matter. The process is as follows: Figure 4As shown in the figure. A grayscale histogram analysis was performed on this figure to obtain the distance distribution from pores to particulate organic matter, as shown below. Figure 6 As shown.

[0083] S6: Since the pore structure and particulate organic matter are independent in the image, the intersection cannot be obtained directly through mathematical operations (Min). Moreover, the Min operation can only obtain a part of the particulate organic matter fragment, not the complete fragment. Therefore, the connected pore image is first dilated by one voxel. Mathematical operations (Min) are then performed on the dilated connected pore image and the particulate organic matter image to obtain the intersection, which is only the edge contour of the particulate organic matter. Then, a three-dimensional morphological reconstruction (Plugins-MorphoLibJ-Filtering-MorphologicalReconstruction3D) is performed on the contour and the particulate organic matter image, where the S61 image is used as the marker image and the particulate organic matter image is used as the mask image. Finally, a complete particulate organic matter image connected to the connected pore structure is obtained. The same steps are followed to obtain a complete particulate organic matter image connected to isolated pores. If a particulate organic matter fragment is connected to both connected pores and isolated pores, it is classified as particulate organic matter connected to connected pores. The Min operation was used to calculate the particulate organic matter connected to connected pore structures and the particulate organic matter connected to isolated pores, obtaining the particulate organic matter connected to both connected and isolated pores. The particulate organic matter connected to isolated pores was then subtracted from the particulate organic matter connected to isolated pores to obtain the particulate organic matter truly connected to isolated pores. By subtracting the particulate organic matter images connected to connected pore structures and those connected to isolated pores from the particulate organic matter images, particulate organic matter images distributed in the soil matrix were obtained. The morphological characteristics of particulate organic matter with different distributions were calculated, as shown in Table 2. Figure 7 and Figure 8 These represent the two-dimensional and three-dimensional distributions of particulate organic matter in the soil.

[0084] S7: Based on temperature response characteristics, particulate organic matter is classified into heat-sensitive POM and heat-stable POM;

[0085] After completing the analyses in S1-S6 above, controlled heating CT scans were performed on the same sample. The soil sample was placed in a CT scanning device equipped with an in-situ heating stage and gradually heated from room temperature (25℃) to 200℃ at a constant heating rate of 5℃ / min. During the heating process, CT scans were performed at eight temperature nodes: 25℃, 50℃, 75℃, 100℃, 125℃, 150℃, 175℃, and 200℃. At each temperature node, the sample was held for 8 minutes to allow the internal temperature to reach uniform equilibrium before scanning. The scanning parameters were kept consistent: voltage 100kV and current 100μA to ensure the comparability of grayscale values ​​of CT images at different temperature nodes.

[0086] For the CT images at each temperature node, particulate organic matter was identified and segmented according to the S2 method. Using the particulate organic matter image at 25℃ as a reference, the CT images at each temperature node were spatially aligned with the reference image using three-dimensional image registration technology (the "LinearStackAlignmentwithSIFT" plugin in Fiji software) to correct for minor displacements that may be caused by thermal expansion. Then, for each fragment identified as particulate organic matter, the "LabelSizeFiltering" and "AnalyzeRegions3D" functions of the MorphoLibJ plugin were used to extract its average grayscale value at each temperature node, and the response curve of grayscale value with temperature change was plotted.

[0087] The grayscale value decay characteristic parameters of each particulate organic matter fragment were calculated, including: initial grayscale value (at 25℃), final grayscale value (at 200℃), total grayscale value decay rate, and grayscale value decay half-temperature point. According to the discrimination criteria, a total grayscale value decay rate greater than 20% is classified as heat-sensitive POM (corresponding to fresh POM), and less than or equal to 20% is classified as heat-stable POM (corresponding to humified old POM). All particulate organic matter fragments in the soil column sample of Example 1 were classified.

[0088] Analysis using the methods described above revealed 1235 particulate organic matter fragments in the soil column sample of Example 1. Among these, 482 fragments were identified as heat-sensitive POM, accounting for approximately 38.7% of the total volume; and 753 fragments were identified as heat-stable POM, accounting for approximately 61.3% of the total volume. The average total attenuation rate of grayscale value for heat-sensitive POM was 34.2% ± 6.1%, while the average total attenuation rate for heat-stable POM was 9.5% ± 3.8%.

[0089] Following the method in S5, distance distribution maps of heat-sensitive POM to the pore surface and heat-stable POM to the pore surface were generated respectively. The results showed that the average distance of heat-sensitive POM to the pore surface was 158 μm, which was significantly smaller than the average distance of heat-stable POM to the pore surface (537 μm). This indicates that heat-sensitive POM (fresh POM) is more distributed near the pore surface, while heat-stable POM (humicated old POM) is distributed in the soil matrix further away from the pore surface.

[0090] Following the method in S6, heat-sensitive POM connected to connected pores and heat-stable POM connected to isolated pores were extracted. The results showed that among POM connected to connected pores, heat-sensitive POM accounted for as high as 71.3%, while heat-stable POM accounted for only 28.7%. Among POM connected to isolated pores, heat-sensitive POM accounted for 45.6%, while heat-stable POM accounted for 54.4%. Among POM distributed within the soil matrix, heat-sensitive POM accounted for only 12.1%, while heat-stable POM accounted for as high as 87.9%. These results indicate that fresh POM (heat-sensitive) is mainly found in well-connected macroporous systems, which is conducive to microbial contact and decomposition. Humic old POM (heat-stable) is more often trapped in isolated pores or within the soil matrix, receiving stronger physical protection, and therefore has higher stability and a slower turnover rate.

[0091] This embodiment also verifies the consistency between the temperature response classification method and the traditional morphological classification method. Existing technologies classify POMs in the same sample based on morphological characteristics such as volume, clumpiness, compactness, sphericity, and plate-like properties. The results show that the agreement between the two classification methods is 86.5%, indicating that the temperature response classification method and the morphological classification method have good complementarity and mutual verification. However, it is worth noting that for POMs with a high degree of decomposition and indistinct morphological features but still retaining certain thermal stability characteristics, the temperature response classification method can effectively identify them, while the morphological classification method has a classification bias of approximately 13.5%. This result indicates that the temperature response classification method has a unique advantage in distinguishing older POMs with a high degree of humification.

[0092] This invention uses particulate organic matter and pore structure as examples to achieve quantitative calculation of their relative spatial distribution. In the fields of soil science and botany, particulate organic matter can be replaced by roots, or pores can be focused on biological pores, thereby analyzing the pore distribution around roots or the root distribution around biological pores, revealing the root-pore interaction mechanism in crop growth. Through morphological reconstruction, the content of particulate organic matter connected to pores can be directly quantified; further, by replacing connected pores with top-connecting pores or permeable pores, the specific influence mechanism of different pore types on particulate organic matter accumulation and root growth can be explored.

[0093] Example 2:

[0094] In this embodiment, topsoil samples (0-10cm) were randomly collected from the tillage and fertilization experimental base of Longkang Farm in Huaiyuan County, Anhui Province. The samples were air-dried naturally, and one aggregate with a diameter of 3-5mm was selected for CT scanning. The method for calculating the relative spatial distribution of particulate organic matter and pore structure in CT images in this embodiment includes the following steps:

[0095] S1: The aggregate samples were scanned using a micro-CT scanner (v|tome|xm300) at a voltage of 100kV and a current of 100μA. Each sample was rotated 360° uniformly on the sample stage, resulting in 1501 images. Images were reconstructed using Datos|x2Rec software and exported as a TIFF sequence. The images were then imported into ImageJ / Fiji software, where Adjust-Brightness / Contrast was used to adjust brightness and contrast. A median filter with a radius of 2 voxels was used for noise reduction, and an anti-sharpening mask filter was used to reduce local volume effects. The images were then saved as a TIFF sequence. Next, the images were imported into VGStudioMAX2022 software, where the adaptive rectangle tool was used to crop the aggregate boundaries, with an isopleth set to 128 and a depth set to 2. The aggregate was then selected as the Region of Interest (ROI). The ROI was inverted and filled with white pixels. The exported image yielded an aggregate image with boundary interference removed, which was then used as the entire ROI for analysis.

[0096] S2: Open the Pixel Classification workflow in the Ilastik software, select Add a Single 3D / 4D Volume from Sequence, import the image from S1, and select pixel features including grayscale value, edge information, and texture features, with Gaussian smoothing standard deviations of δ=0.3, 0.7, and 1.0, respectively. Classify the original image into soil matrix, pores, and particulate organic matter, and add three labels accordingly, each corresponding to a pixel category. Then, use the brush tool to add annotations to each pixel category and perform classification training (LiveUpdate). If misclassification occurs, re-label the pixels using the brush or eraser function until the training result is accurate. The exported image data type is Integer8bit, and the image format is TiffSequence.

[0097] The segmented image from Ilastik was imported into Fiji software. Color-SplitChannels was used to split the RGB channels of the image into soil matrix, pores, and granular organic matter. Image sequences containing granular organic matter and pores were retained and filtered using a median filter with a radius of 2 pixels. The image grayscale values ​​were then inverted. The image was binarized using the Sauvola algorithm in AutoLocalThreshold, with default parameter settings. After binarization, the black areas in the image were larger than one voxel (6×6×6μm). 3 The particulate organic matter or pores are sieved. The particulate organic matter is sieved to obtain a sieve volume greater than 50×50×50μm. 3 For the particulate organic matter, the ConnectedComponentsLabeling function of the MorphoLibJ plugin was used to add labels to the black pixels. Then, LabelSizeFiltering was used to screen out the particulate organic matter with a volume greater than 579 voxels and convert it into a binary image. Figure 9 Images of pores and granular organic matter obtained after image segmentation of grayscale images are listed.

[0098] S3: Calculate the basic characteristics of the pores using the AnalyzeRegions3D function of the MorphoLibJ plugin: volume, surface area, average width, sphericity, and Euler number; as shown in Table 3.

[0099] Table 3 Pore structure characteristic parameters at the aggregate scale

[0100]

[0101] Separating Connected Pores from Isolated Pores: In this embodiment 2, the second connected pore definition method described in the specification is used for separation, that is, pores connected to the outside of the aggregate are defined as connected pores, and internal pores are isolated pores. First, the grayscale image is thresholded (threshold range is set to 0-254). Since the grayscale values ​​of high-density materials such as sand and gravel in the image are high (255), the threshold-adjusted image needs to be filled (Process-Binary-FillHoles) so that the region of interest (ROI) is black (grayscale value is 255), thus obtaining a mask image, such as... Figure 10 As shown in the figure. Then, the mask image is subtracted from the pore image to obtain image 1. The result is then filled with internal isolated pores (FillHole) to obtain image 2. At this point, only the surface connected pores are not filled. Subtracting image 2 from image 1 yields the connected pores. Subtracting the connected pores from the pore image yields the isolated pores. Morphological features are calculated, as shown in Table 3. The three-dimensional distribution of isolated and connected pores is as follows: Figure 11 As shown.

[0102] S4: The basic morphological characteristics of the POM (PoM) were calculated using the MorphoLibJ plugin's AnalyzeRegions3D function: volume, surface area, average width, and sphericity, as shown in Table 4. The particle size distribution and average diameter of the POM were calculated using Bonej's Thickness function. The particle size distribution is shown in Table 4. Figure 12 As shown, the average diameter of POM is 52 μm.

[0103] Table 4. Morphological characteristics of particulate organic matter at the soil column scale

[0104]

[0105] Note: POM connected, POM isolated, and POM matrix refer to particulate organic matter connected to connected pores, connected to isolated pores, and distributed in the soil matrix, respectively.

[0106] S5: Calculate the distance distribution from particulate organic matter to the pore surface: Subtract the pore image from the mask image, and perform Euclidean distance transformation (Plugins-Process-ExactEuclideanDistanceTransform (3D)) on the resulting image to obtain a 3D distance distribution map. This map represents the distance distribution from scene voxels (including soil matrix and particulate organic matter) to the nearest pore surface; the larger the gray value, the farther the distance. Perform gray-level histogram analysis on this distance distribution map to obtain the distance distribution, and calculate the average pore distance, which is the weighted average distance from each non-pore voxel to the nearest pore voxel, which is 81 μm. Generally, the higher the pore connectivity, the smaller the average pore distance. Divide the gray values ​​of the particulate organic matter image by 255 (Process-Math-Divide) to make all particulate organic matter gray values ​​1. Multiply the normalized particulate organic matter image with the 3D distance distribution map to obtain the distance distribution map of particulate organic matter relative to the pore surface. The process is as follows: Figure 13 As shown in the figure. A grayscale histogram analysis was performed on this figure to obtain the distance distribution from the particulate organic matter to the pores, as shown below. Figure 14 As shown.

[0107] Calculate the distance distribution from pores to the surface of particulate organic matter: Subtract the particulate organic matter image from the mask image, perform Euclidean distance transformation (Plugins-Process-ExactEuclideanDistanceTransform (3D)) on the resulting image to obtain a 3D distance distribution map. Analyze the grayscale histogram of this distance distribution map to obtain the distance distribution, allowing for the calculation of the average particulate organic matter distance, which is the weighted average distance from each non-porous voxel to the nearest pore voxel, and is 74 μm. Generally, the higher the particulate organic matter content, the smaller the average particulate organic matter distance. Divide the grayscale values ​​of the pore image by 255 (Process-Math-Divide) to make all pore grayscale values ​​1. Multiply the normalized pore image with the 3D distance distribution map to obtain the distance distribution map of pores relative to the surface of particulate organic matter. The process is as follows: Figure 13 As shown in the figure. A grayscale histogram analysis was performed on this figure to obtain the distance distribution from pores to particulate organic matter, as shown below. Figure 15 As shown.

[0108] S6: Since the pore structure and particulate organic matter are independent in the image, the intersection cannot be obtained directly through mathematical operations (Min). Moreover, the Min operation can only obtain a part of the particulate organic matter fragment, not the complete fragment. Therefore, the connected pore image is first dilated by one voxel. Mathematical operations (Min) are then performed on the dilated connected pore image and the particulate organic matter image to obtain the intersection, which is only the edge contour of the particulate organic matter. Then, a three-dimensional morphological reconstruction (Plugins-MorphoLibJ-Filtering-MorphologicalReconstruction3D) is performed on the contour and the particulate organic matter image, where the S61 image is used as the marker image and the particulate organic matter image is used as the mask image. Finally, a complete particulate organic matter image connected to the connected pore structure is obtained. The same steps are followed to obtain a complete particulate organic matter image connected to isolated pores. If a particulate organic matter fragment is connected to both connected pores and isolated pores, it is classified as particulate organic matter connected to connected pores. The Min operation was used to calculate the particulate organic matter connected to connected pore structures and the particulate organic matter connected to isolated pores, obtaining the particulate organic matter connected to both connected and isolated pores. The particulate organic matter connected to isolated pores was then subtracted from the particulate organic matter connected to isolated pores to obtain the particulate organic matter truly connected to isolated pores. By subtracting the particulate organic matter images connected to connected pore structures and those connected to isolated pores from the particulate organic matter images, particulate organic matter images distributed in the soil matrix were obtained. The morphological characteristics of particulate organic matter with different distributions were calculated, as shown in Table 4. Figure 16 and Figure 17 The images show the two-dimensional and three-dimensional distribution of particulate organic matter in the soil. It can be observed that particulate organic matter is mainly enclosed within the isolated pores of aggregates, thus receiving physical protection from the pore structure of the aggregates.

[0109] S7: Based on temperature response characteristics, particulate organic matter is classified into heat-sensitive POM and heat-stable POM;

[0110] Controlled heating CT scan analysis was also performed on the aggregate samples in Example 2. Since the aggregate samples are small in volume (3-5 mm in diameter) and have more uniform and rapid heat conduction, the holding time was adjusted to 5 minutes to achieve uniform temperature equilibrium inside the sample. The heating program was the same as in Example 1: the temperature was gradually increased from 25°C to 200°C at a constant heating rate of 5°C / min. CT scans were performed at eight temperature nodes: 25°C, 50°C, 75°C, 100°C, 125°C, 150°C, 175°C and 200°C, and the scanning parameters remained consistent.

[0111] For the CT images at each temperature node, particulate organic matter was identified and segmented according to the method in S2. After image registration and spatial alignment using the particulate organic matter image at 25℃ as a reference, the average gray value of each particulate organic matter fragment at each temperature node was extracted, gray value-temperature response curves were plotted, and gray value attenuation characteristic parameters were calculated. The particulate organic matter fragments in the aggregate sample were classified according to the same discrimination criteria as in Example 1 (a total gray value attenuation rate greater than 20% indicates heat-sensitive POM, and less than or equal to 20% indicates heat-stable POM).

[0112] Analysis using the methods described above identified 687 particulate organic fragments in the aggregated sample of Example 2. Among them, 294 fragments were identified as heat-sensitive POM, accounting for approximately 41.2% of the total volume; and 393 fragments were identified as heat-stable POM, accounting for approximately 58.8% of the total volume. The average total attenuation rate of grayscale value for heat-sensitive POM was 36.5% ± 7.2%, and the average total attenuation rate of grayscale value for heat-stable POM was 8.7% ± 3.1%.

[0113] Further spatial analysis revealed that, at the aggregate scale, the average distance from heat-sensitive POM to the pore surface was 26 μm, significantly smaller than the average distance from thermally stable POM to the pore surface (98 μm). Regarding its association with pore structure, heat-sensitive POM accounted for 63.8% of the POM connected to interconnected pores; 52.1% of the POM connected to isolated pores; and only 8.7% of the POM distributed within the soil matrix. Compared to the soil column scale results of Example 1, At the aggregate scale, the proportion of heat-sensitive POMs in isolated pores was higher (52.1% vs 45.6%), reflecting the more significant role of the microporous structure inside the aggregates in trapping and protecting fresh organic matter. At the same time, at both scales, heat-sensitive POMs were preferentially distributed near the pore surface and in connected pore systems, while heat-stable POMs tended to be distributed in the soil matrix far from the pores. This pattern showed good consistency across the two spatial scales, demonstrating the applicability and robustness of the temperature response classification method at different scales.

[0114] Table 5. Summary of temperature response classification and spatial distribution characteristics of particulate organic matter in soil column samples from Example 1 and aggregate samples from Example 2.

[0115]

[0116] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of protection claimed by the present invention. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for calculating the relative spatial distribution of particulate organic matter and pore structure in CT images, characterized in that, Includes the following steps: Step S1: Perform CT scans on soil samples and preprocess the images to obtain preprocessed CT images; Step S2: Based on machine learning, identify and segment particulate organic matter and pore structure from the preprocessed CT images; Step S3: Calculate the morphological characteristics of the pore structure; Step S4: Calculate the morphological characteristics of the particulate organic matter; Step S5: Analyze the relative spatial distribution of the particulate organic matter and the pore structure, including calculating the distance distribution from the particulate organic matter to the surface of the pore structure; as well as Step S6: Analyze and extract the particulate organic matter connected to the pore structure; Step S7: Classify the particulate organic matter based on temperature response characteristics, specifically including: During the CT scan, a controlled heating procedure was performed on the soil sample, and corresponding CT images were acquired at different temperature points. Based on machine learning, the particulate organic matter at different temperature nodes is identified, and the spatial location of particulate organic matter at different temperature nodes is matched by 3D image registration. The gray values ​​of the same organic matter fragment at different temperature nodes were extracted, and gray value-temperature response curves were established. Based on the gray value attenuation characteristic parameters of particulate organic matter fragments, particulate organic matter is classified into heat-sensitive particulate organic matter and heat-stable particulate organic matter. The relative spatial distribution relationship between the heat-sensitive particulate organic matter and the heat-stable particulate organic matter and the pore structure is calculated respectively.

2. The method according to claim 1, characterized in that, Step S2 further includes: Supervised machine learning tools are used to classify each pixel in the preprocessed CT image, with classification labels including at least soil matrix, particulate organic matter, and pores; Based on the classification results, binary images corresponding to particulate organic matter and pore structure are generated respectively.

3. The method according to claim 2, characterized in that, In step S2, the particle organic matter image and pore structure image obtained after classification are further post-processed. The post-processing includes: median filtering, gray value inversion, local threshold binarization, and particle organic matter sieving based on volume threshold.

4. The method according to claim 1, characterized in that, In step S3, the morphological characteristics of the pore structure are calculated, including the volume, surface area, average width, sphericity, Euler number, and pore size distribution of one or more of the following: total pores, interconnected pores, isolated pores, top-connected pores, and permeable pores. The connected pores and isolated pores are classified in any of the following ways: (i) Define the largest pore as a connected pore, and the rest as isolated pores; or (ii) Define pores that are connected to the boundary of the region of interest as connected pores, and define pores that are completely located inside the region of interest as isolated pores.

5. The method according to claim 1, characterized in that, In step S4, the morphological characteristics of particulate organic matter are calculated, including the volume, surface area, average width, sphericity, and particle size distribution of the particulate organic matter.

6. The method according to claim 1, characterized in that, The step S5, which analyzes the relative spatial distribution of particulate organic matter and pore structure, further includes: Generate a three-dimensional distance distribution map representing the distances from the soil matrix and particulate organic matter to the nearest pore surface; Multiplying the three-dimensional distance distribution map with the normalized particulate organic matter image yields a distance distribution map of particulate organic matter relative to the pore surface, which is used to characterize the degree to which particulate organic matter is physically protected by pores.

7. The method according to claim 6, characterized in that, Step S5 further includes: The particulate organic matter and pore structure are interchanged to generate a three-dimensional distance distribution map representing the soil matrix and pores to the nearest particulate organic matter surface. This distance distribution map is then multiplied with the normalized pore image to obtain a distance distribution map of pores relative to the particulate organic matter surface.

8. The method according to claim 1, characterized in that, The analysis and extraction of particulate organic matter connected to the pore structure in step S6 specifically includes: Dilate the pore structure image by one voxel; Mathematical operations were performed on the expanded pore structure image and the particulate organic matter image to obtain the intersection of the two; Using the intersecting portion as a marker image and the original particulate organic matter image as a mask image, three-dimensional morphological reconstruction is performed to extract particulate organic matter that is fully connected to the pore structure.

9. The method according to claim 8, characterized in that, Step S6 further includes: The total pore image was replaced with connected pore images and isolated pore images respectively, and the particulate organic matter connected to connected pores and the particulate organic matter connected to isolated pores were extracted. When the same particle of organic matter is connected to both connected pores and isolated pores, it is classified as being connected to connected pores. The particulate organic matter distributed within the soil matrix is ​​obtained by subtracting the particulate organic matter connected to interconnected pores and the particulate organic matter connected to isolated pores from the total particulate organic matter.

10. The method according to any one of claims 1-9, characterized in that, The method is applied to the analysis of CT images of soil samples at the soil column scale and / or aggregate scale; and the particulate organic matter can be replaced with plant roots, and the pore structure can be replaced with biological pores to analyze the root-pore interaction mechanism.

Images