Quantitative detection method and detection system for soil-rock mixture micro-defects based on CT image multi-threshold segmentation

Abstract

This invention proposes a quantitative detection method and system for microscopic defects in soil-rock mixtures based on multi-threshold segmentation of CT images, belonging to the field of non-destructive testing and digital image processing technology for geotechnical engineering materials. The method includes: firstly, using a fuzzy C-means clustering algorithm to perform multi-threshold segmentation on CT images, resolving the image into four regions based on grayscale features: background, boulders, matrix soil, and microscopic defects; then, based on the material's heterogeneity, classifying microscopic defects into matrix defects and interface defects according to their spatial distribution; extracting the boundary contours of boulders through morphological operations, and combining this with spatial topological analysis of the binary defect image to achieve intelligent defect type discrimination; finally, quantifying the volume proportion parameters of the two types of defects based on pixel statistics. This method overcomes the limitations of traditional homogeneous material defect analysis methods, providing a new technical means for studying the damage evolution of multiphase structures in soil-rock mixtures.

Description

Technical Field

[0001] This invention relates to the field of nondestructive testing and digital image processing technology for geotechnical engineering materials, specifically to a method and system for quantitative detection of microscopic defects in soil-rock mixtures based on multi-threshold segmentation of CT images. Background Technology

[0002] Soil-rock mixtures, as typical heterogeneous materials widely found in landslides, sedimentary layers, and other geological engineering applications, exhibit mechanical property degradation closely related to the evolution of internal microscopic defects (such as pores and fissures). Traditional geotechnical testing methods often rely on physical and mechanical tests or two-dimensional slice observation, which have limitations such as damaging sample integrity and difficulty in three-dimensionally visualizing defect distribution. Although computed tomography (CT) technology has made non-destructive testing possible, when dealing with multiphase media like soil-rock mixtures, conventional image segmentation methods (such as single thresholding and region growing methods) are prone to misjudgment or missed detection of defects due to issues such as overlapping gray levels of material components and blurred interfaces. This makes it difficult to achieve accurate separation of rocks, matrix, and defects, and even more difficult to distinguish the differentiated effects of interface defects and matrix defects on mechanical properties.

[0003] Existing patents and literature on CT image-based defect analysis primarily focus on homogeneous materials (such as concrete and rock), lacking dedicated methods for heterogeneous soil-rock mixtures. For example, traditional clustering algorithms fail to consider the constraints of block boundary morphology on defect classification, resulting in insufficient accuracy in interface defect identification. Furthermore, defect quantification methods relying solely on grayscale statistics struggle to establish quantitative correlations between microscopic defect parameters and macroscopic indicators such as modulus and strength. Moreover, current technologies often neglect the impact of heterogeneous spatial distribution of defects on engineering stability, failing to meet the needs of landslide monitoring, roadbed assessment, and other scenarios requiring defect type identification and damage evolution analysis. Therefore, there is an urgent need to develop a defect detection method integrating multimodal image processing to address the adaptability challenges in heterogeneous material defect analysis. Summary of the Invention

[0004] The purpose of this invention is to propose a quantitative detection method and system for microscopic defects in soil-rock mixtures based on multi-threshold segmentation of CT images. By integrating fuzzy clustering algorithms, morphological operations, and spatial topology analysis techniques, it achieves innovation across the entire process from image analysis to defect classification and assessment.

[0005] The technical solution adopted in this invention is as follows: Firstly, this invention proposes a method for quantitative detection of microscopic defects in soil-rock mixtures based on multi-threshold segmentation of CT images, the method comprising:

[0006] Obtain the original CT image of the soil-rock mixture and extract its grayscale data;

[0007] The grayscale data is segmented using a fuzzy C-means clustering algorithm to determine three grayscale thresholds: Th1, Th2, and Th3. Th1 is the threshold for distinguishing between the background and the target object, Th2 is the threshold for distinguishing between boulders and the matrix soil, and Th3 is the threshold for distinguishing between the matrix soil and microscopic defects.

[0008] Based on the three grayscale thresholds, the original CT image is segmented into four regions: background, boulders, matrix soil, and microscopic defects, and a binary image of the defects is generated.

[0009] Morphological processing is performed on the rock area, and the rock boundary contour is extracted by dilation-erosion operation to generate a binary image of the rock boundary;

[0010] Spatial topological analysis is performed on the binary image of the defect and the binary image of the boulder boundary. The interface defect and the matrix defect are separated by the image subtraction operation. The interface defect is the defect region that overlaps with the boulder boundary, and the matrix defect is the defect region that is independently distributed in the matrix soil.

[0011] The pixel percentages of the interface defects and matrix defects are statistically analyzed, their volume percentages in the overall image of the soil-rock mixture are calculated, and quantitative detection results are output.

[0012] As a further improvement of the present invention, the objective function of the fuzzy C-means clustering algorithm is defined as:

[0013]

[0014] Where N is the number of data points, C is the number of clusters, and μ ij For pixels x i The membership degree of class j, c j is the cluster center, and m is the fuzzy weight coefficient. For a given data point, the sum of its membership degree to each class is 1.

[0015] As a further improvement of the present invention, the kernel function of the morphologically treated expansion-corrosion operation is a circular structural element with a radius that is 0.1-0.2 times the average particle size of the stone.

[0016] As a further improvement of the present invention, the spatial topology analysis includes:

[0017] Perform a logical AND operation between the binary image of the stone boundary and the binary image of the defect, and extract the overlapping area as the interface defect;

[0018] Subtract the interface defect region from the binary image of the defect, and the remaining part is defined as the matrix defect.

[0019] As a further improvement of the present invention, the formula for calculating the volume ratio is as follows:

[0020]

[0021] Where, N defect N represents the number of pixels in the defect area. total This represents the total number of pixels in the effective area of ​​the soil-rock mixture.

[0022] Secondly, this invention also proposes a quantitative detection system for microscopic defects in soil-rock mixtures based on multi-threshold segmentation of CT images, the detection system comprising:

[0023] The image acquisition module is used to acquire the original CT image of the soil-rock mixture and extract its grayscale data;

[0024] The image processing module is used to perform multi-threshold segmentation on the grayscale data using a fuzzy C-means clustering algorithm to determine three grayscale thresholds Th1, Th2, and Th3, where Th1 is the threshold for distinguishing between the background and the target object, Th2 is the threshold for distinguishing between boulders and the matrix soil, and Th3 is the threshold for distinguishing between the matrix soil and microscopic defects; and to segment the original CT image into four regions—background, boulders, matrix soil, and microscopic defects—based on the three grayscale thresholds, and generate a binary image of the defects; and to perform morphological processing on the boulder region, extracting the boulder boundary contour through a dilatation-erosion operation to generate a binary image of the boulder boundary.

[0025] The defect classification module is used to perform spatial topological analysis on the binary image of the defect and the binary image of the rock boundary, and to separate the interface defect and the matrix defect through the image subtraction operation. The interface defect is the defect area that overlaps with the rock boundary, and the matrix defect is the defect area that is independently distributed in the matrix soil.

[0026] The statistical output module is used to statistically analyze the pixel proportion of the interface defects and matrix defects, calculate their volume proportion in the overall image of the soil-rock mixture, and output quantitative detection results.

[0027] Thirdly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed, implements the steps of any of the methods described above.

[0028] Compared with the prior art, the present invention has the following technical effects:

[0029] (1) This invention uses the fuzzy C-means clustering algorithm to perform multi-threshold segmentation on grayscale data of CT images, so that each pixel has the highest membership degree in its corresponding category, and the segmentation result is accurate and reliable, accurately dividing the grayscale data into four types of regions: background, boulders, matrix soil and micro-defects.

[0030] (2) Based on the different specific locations of defects, the present invention performs morphological analysis and spatial topological operations on CT images to further distinguish the microscopic defects of soil-rock mixtures into matrix defects and interface defects, which can more effectively reflect different microscopic damage mechanisms.

[0031] (3) The present invention is based on pixel statistical quantification of the volume ratio of two types of defects, which can be used as the basis for the evolution of related macroscopic mechanical properties.

[0032] (4) This method breaks through the limitations of the traditional framework for analyzing defects in homogeneous materials and significantly improves the accuracy and engineering applicability of defect detection in heterogeneous materials. Attached Figure Description

[0033] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0034] Figure 1 This is an image-based flowchart of the intelligent discrimination and quantitative detection method for microscopic defects in soil-rock mixtures based on multi-threshold segmentation of CT images according to the present invention.

[0035] Figure 2 This is a schematic diagram of the defect classification results of the freeze-thawed soil-rock mixture sample in the embodiment. Detailed Implementation

[0036] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention are within the scope of protection of the present invention.

[0037]

Example 1

[0038] This invention proposes a method for quantitative detection of microscopic defects in soil-rock mixtures based on multi-threshold segmentation of CT images, comprising:

[0039] Obtain the original CT image of the soil-rock mixture and extract its grayscale data.

[0040] Fuzzy C-means clustering algorithm is used to perform multi-threshold segmentation on grayscale data, determining three grayscale thresholds Th1, Th2, and Th3. Th1 is the threshold for distinguishing between background and target objects, Th2 is the threshold for distinguishing between boulders and matrix soil, and Th3 is the threshold for distinguishing between matrix soil and microscopic defects. Specifically, the objective function of the fuzzy C-means clustering algorithm is defined as:

[0041]

[0042] Where N is the number of data points, C is the number of clusters, and μij For pixels x i The membership degree of class j, c j is the cluster center, and m is the fuzzy weight coefficient. For a given data point, the sum of its membership degree to each class is 1.

[0043] The original CT image is segmented into four regions—background, boulders, matrix soil, and micro-defects—based on three grayscale thresholds, and a binary image of the defects is generated.

[0044] Morphological processing was performed on the rocky area, and the rock boundary contours were extracted through dilation-erosion operations to generate binary images of the rock boundaries. Specifically, the kernel function of the dilation-erosion operation in the morphological processing was a circular structuring element, the ratio of its radius to the average grain size of the rocks being 0.1-0.2.

[0045] Spatial topological analysis was performed on the binary images of defects and the boulder boundary. Interface defects and matrix defects were separated by image subtraction. Interface defects are the defect regions overlapping with the boulder boundary, while matrix defects are the defect regions independently distributed within the matrix soil. The spatial topological analysis included: performing a logical AND operation between the binary images of the boulder boundary and the defects to extract the overlapping region as the interface defect; and subtracting the interface defect region from the binary image of the defects, with the remaining portion defined as the matrix defect.

[0046] The pixel percentages of interface defects and matrix defects are statistically analyzed, and their volume proportions relative to the overall image of the soil-rock mixture are calculated, outputting quantitative detection results. Specifically, the formula for calculating the volume proportion is:

[0047]

[0048] Where, N defect N represents the number of pixels in the defect area. total This represents the total number of pixels in the effective area of ​​the soil-rock mixture.

[0049] Figure 1 This invention demonstrates the complete process and key steps of the quantitative detection method for microscopic defects in soil-rock mixtures based on multi-threshold segmentation of CT images, specifically including:

[0050] 1. Input of original CT images (a): Three-dimensional CT image of soil-rock mixture sample after freeze-thaw cycle, grayscale range is 0-255.

[0051] 2. Gray-level frequency distribution and threshold segmentation of image (bc):

[0052] The horizontal axis is the gray value (0 - 255), and the vertical axis is the pixel frequency; the three - level thresholds (Th1, Th2, Th3) are determined by the fuzzy C - means clustering algorithm, and the image is segmented into four types of regions: background (gray value ≥ Th1), block stones (Th2 ≤ gray value < Th1), matrix soil (Th3 ≤ gray value < Th2), and meso - scale defects (gray value < Th3).

[0053] 3. Morphological processing and defect classification:

[0054] The binary image of block stones (d) is processed by dilation - erosion operations to extract the boundary (e);

[0055] The logical operation is performed between the binary image of defects (g) and the boundary of block stones to separate the interface defects (f) and matrix defects (h);

[0056] 4. Pixel statistics and result output:

[0057] Quantitatively display the proportion of matrix defects and the proportion of interface defects.

[0058] Figure 2 The defect classification results of the soil - rock mixture specimens after 6 freeze - thaw cycles in the embodiment are shown. Among them, Figure 2 (a) is the distribution map of interface defects. The red - highlighted area is the defects (pores / fissures) at the soil - rock interface, which are extracted by the logical "AND" operation between the binary image of the block - stone boundary and the binary image of defects, reflecting the damage concentration phenomenon in the contact zone between block stones and matrix soil. Figure 2 (b) is the distribution map of matrix defects. The blue area is the independent defects inside the matrix soil, which are obtained by subtracting the interface defects from the total defect area, characterizing the deterioration of the internal structure of the soil caused by freeze - thaw cycles.

[0059]

Embodiment 2

[0060] The present invention proposes a meso - scale defect quantitative detection system for soil - rock mixtures based on multi - threshold segmentation of CT images, including:

[0061] An image acquisition module, used to obtain the original CT image of the soil - rock mixture and extract its gray - scale data;

[0062] An image processing module, used to perform multi - threshold segmentation on the gray - scale data by the fuzzy C - means clustering algorithm to determine three gray - scale thresholds Th1, Th2, and Th3. Among them, Th1 is the threshold for distinguishing the background from the target object, Th2 is the threshold for distinguishing block stones from matrix soil, and Th3 is the threshold for distinguishing matrix soil from meso - scale defects; and segment the original CT image into four regions: background, block stones, matrix soil, and meso - scale defects according to the three gray - scale thresholds, and generate a binary image of defects; and perform morphological processing on the block - stone region, and extract the boundary contour of block stones through dilation - erosion operations to generate a binary image of the block - stone boundary;

[0063] The defect classification module is used to perform spatial topological analysis on the binary image of defects and the binary image of the rock boundary. It separates interface defects and matrix defects through image subtraction. Interface defects are defect areas that overlap with the rock boundary, while matrix defects are defect areas that are independently distributed inside the matrix soil.

[0064] The statistical output module is used to statistically analyze the pixel proportion of interface defects and matrix defects, calculate their volume proportion in the overall image of the soil-rock mixture, and output quantitative detection results.

[0065]

Example 3

[0066] This invention provides a computer-readable storage medium storing a computer program thereon, which, when executed, implements the steps of any of the above methods.

[0067] This method targets the original CT images of soil-rock mixtures. By fusing fuzzy clustering algorithms, morphological operations, and spatial topology analysis, it separates interface defects and matrix defects. Then, it counts the pixel proportions of interface defects and matrix defects and calculates their volume proportions in the overall image of the soil-rock mixture. This achieves a complete innovation from image analysis to defect classification and assessment.

[0068] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited thereto. Various changes that can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention are all within the protection scope of the claims of the present invention.

Claims

1. A method for quantitative detection of microscopic defects in soil-rock mixtures based on multi-threshold segmentation of CT images, characterized in that, The method includes: Obtain the original CT image of the soil-rock mixture and extract its grayscale data; The grayscale data is segmented using a fuzzy C-means clustering algorithm to determine three grayscale thresholds: Th1, Th2, and Th3. Th1 is the threshold for distinguishing between the background and the target object, Th2 is the threshold for distinguishing between boulders and the matrix soil, and Th3 is the threshold for distinguishing between the matrix soil and microscopic defects. Based on the three grayscale thresholds, the original CT image is segmented into four regions: background, boulders, matrix soil, and microscopic defects, and a binary image of the defects is generated. Morphological processing is performed on the rock area, and the rock boundary contour is extracted by dilation-erosion operation to generate a binary image of the rock boundary; Spatial topological analysis is performed on the binary image of the defects and the binary image of the boulder boundary. Interface defects and matrix defects are separated through image subtraction. Interface defects are defect regions overlapping with the boulder boundary, while matrix defects are defect regions independently distributed within the matrix soil. The spatial topological analysis includes: Perform a logical AND operation between the binary image of the block boundary and the binary image of the defect to extract the overlapping area as the interface defect; subtract the interface defect area from the binary image of the defect, and define the remaining part as the matrix defect. The pixel percentages of the interface defects and matrix defects are statistically analyzed, their volume percentages in the overall image of the soil-rock mixture are calculated, and quantitative detection results are output.

2. The method for quantitative detection of microscopic defects in soil-rock mixtures based on multi-threshold segmentation of CT images according to claim 1, characterized in that, The objective function of the fuzzy C-means clustering algorithm is defined as: ； in, For the number of data points, The number of clusters, For pixels Belongs to the Membership degree of a class As cluster center, For a given data point, the sum of its membership degrees to each category is 1.

3. The method for quantitative detection of microscopic defects in soil-rock mixtures based on multi-threshold segmentation of CT images according to claim 1, characterized in that, The kernel function of the morphologically treated expansion-corrosion operation is a circular structural element with a radius that is 0.1-0.2 times the average particle size of the stone.

4. The method for quantitative detection of microscopic defects in soil-rock mixtures based on multi-threshold segmentation of CT images according to claim 1, characterized in that, The formula for calculating the volume ratio is: ； in, This represents the number of pixels in the defect area. This represents the total number of pixels in the effective area of ​​the soil-rock mixture.

5. A quantitative detection system for microscopic defects in soil-rock mixtures based on multi-threshold segmentation of CT images, characterized in that, The detection system includes The image acquisition module is used to acquire the original CT image of the soil-rock mixture and extract its grayscale data; The image processing module is used to perform multi-threshold segmentation on the grayscale data using a fuzzy C-means clustering algorithm to determine three grayscale thresholds Th1, Th2, and Th3, where Th1 is the threshold for distinguishing between the background and the target object, Th2 is the threshold for distinguishing between boulders and the matrix soil, and Th3 is the threshold for distinguishing between the matrix soil and microscopic defects; and to segment the original CT image into four regions—background, boulders, matrix soil, and microscopic defects—based on the three grayscale thresholds, and generate a binary image of the defects; and to perform morphological processing on the boulder region, extracting the boulder boundary contour through a dilatation-erosion operation to generate a binary image of the boulder boundary. The defect classification module is used to perform spatial topological analysis on the binary image of the defects and the binary image of the rock boundary. It separates interface defects and matrix defects through image subtraction. Interface defects are defect regions that overlap with the rock boundary, while matrix defects are defect regions independently distributed within the matrix soil. The spatial topological analysis includes: performing a logical AND operation on the binary image of the rock boundary and the binary image of the defects to extract the overlapping region as the interface defect; subtracting the interface defect region from the binary image of the defects, and defining the remaining part as the matrix defect. The statistical output module is used to statistically analyze the pixel proportion of the interface defects and matrix defects, calculate their volume proportion in the overall image of the soil-rock mixture, and output quantitative detection results.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed, it implements the steps of the method as described in any one of claims 1-4.

Images