~黄土是一种具有多孔介质的岩土工程材料，准确表征黄土的孔隙结构对研究黄土的渗透性具有极其重要的意义。目前，对黄土孔隙进行的研究多集中在孔隙的孔径、长度等定量参数。黄土中的垂直孔隙较为发育，具有明显的矢量特征，需要对其矢量化特征进行分析。本文在MATLAB中编程相关的数学算法，建立了一套CT图像处理和三维孔隙模型定量表征的高效工作流程。在验证该流程的准确性后，使用参数定量对比了取自绥德、洛川、西安三个地区黄土试样的3D孔隙结构。基于模型获取参数，利用与CT相结合的结构张量算法对黄土试样进行矢量化表征对比，评价其孔隙分布空间特性。得出主要成果如下：
（1）利用CT技术获取绥德、洛川、西安、洛川-1四个试样的二维切片图像，基于MATLAB的算法对四个试样进行了定性分析和定量表征，结果表明本文开发的代码能够有效的表征黄土孔隙结构。
（2）因为试样的尺度不同，其孔隙结构也呈现出典型的不同，大尺寸圆柱形试样的孔隙结构为管状孔隙，且垂直方向大孔隙发育明显。这些从顶部到底部延伸的孔隙具有良好的连通性，对水力传导性的影响更大，并通过渗流模拟进行了验证。相反小尺寸洛川-1试样孔隙结构呈海绵状结构。
（3）通过定量分析黄土试样的孔隙结构，绥德、洛川和西安三个试样的孔隙率、孔隙个数、成圆率均自上而下呈逐渐减小趋势。试样的孔径分布曲线均呈偏态的单峰分布，符合高斯分布。相较于洛川试样，洛川-1试样可以观察到更细小的孔隙，洛川-1试样的孔隙个数和孔隙率远大于洛川试样，观察的孔隙成圆率更加稳定。洛川-1小尺度试样的CT扫描结果对洛川黄土大孔隙的孔隙分布曲线进行了补充，将宏微观尺度进行连接，实现多尺度的孔隙结构研究。
（4）绥德、洛川、西安和洛川-1试样的结构张量系数分别为5.2706、0.8602、0.0954、0.0062。结构张量系数越大各向异性越强，依据结构张量系数可以看出绥德试样的结构张量系数最大，其孔隙结构的各向异性最强，这与分形维数的表征结果一致，绥德试样分形维数最大，为2.9。这表明结构张量能够准确的描述黄土孔隙结构的分布情况。
（5）在基于孔隙结构的水土特征曲线预测中，微米级CT的洛川-1试样与SEM扫描试样的预测进气值相近，这是因为两者分辨率相近，但洛川-1试样的孔隙更小，其曲线呈现更缓分布。绥德、洛川、西安试样对于SEM预测水土特征曲线拥有更小的进气值，因为其平均孔径均大于SEM结果，所以曲线相对SEM图像曲线呈现更陡的趋势。

~Loess is a kind of geotechnical engineering material with porous media. Accurate characterization of the pore structure of loess is extremely important for studying the permeability of loess. At present, the research on the pores of loess mostly focuses on quantitative parameters such as the pore diameter and length. The vertical pores in loess are relatively developed and have obvious vector characteristics, which need to be analyzed. In this paper, related mathematical algorithms are programmed in MATLAB, and a set of efficient workflows for CT image processing and quantitative characterization of three-dimensional pore models are established. After verifying the accuracy of the process, parameters were used to quantitatively compare the 3D pore structures of loess samples taken from Suide, Luochuan, and Xi'an. Based on the parameters obtained from the model, the structural tensor algorithm combined with CT is used to perform vectorized characterization and comparison of the loess samples, and evaluate the spatial characteristics of the pore distribution. The main results are as follows:
(1) Using CT technology to obtain two-dimensional slice images of four samples from Suide, Luochuan, Xi'an, and Luochuan-1, the four samples were qualitatively analyzed and quantitatively characterized based on the algorithm of MATLAB. The results show that the development of this article The code can effectively characterize the pore structure of loess.
(2) Because of the different scales of the samples, the pore structure also shows typical differences. The pore structure of the large-size cylindrical sample is tubular pores, and the large pores are obviously developed in the vertical direction. These pores extending from the top to the bottom have good connectivity and have a greater impact on hydraulic conductivity, and are verified by seepage simulation. On the contrary, the pore structure of the small-sized Luochuan-1 sample is a sponge-like structure.
(3) Through the quantitative analysis of the pore structure of the loess samples, the porosity, number of pores and roundness of the three samples from Suide, Luochuan and Xi'an all show a gradually decreasing trend from top to bottom. The pore size distribution curves of the samples all showed a skewed unimodal distribution, conforming to the Gaussian distribution. Compared with the Luochuan sample, the Luochuan-1 sample can observe finer pores. The number of pores and porosity of the Luochuan-1 sample is much larger than that of the Luochuan sample, and the observed pore rounding rate is more stable. The CT scan results of the Luochuan-1 small-scale sample complement the pore distribution curve of the large pores in Luochuan loess, and connect the macro and micro scales to realize multi-scale pore structure research.
(4) The structural tensor coefficients of the samples from Suide, Luochuan, Xi'an and Luochuan-1 are 5.2706, 0.8602, 0.0954, 0.0062, respectively. The larger the structure tensor coefficient, the stronger the anisotropy. According to the structure tensor coefficient, it can be seen that the structure tensor coefficient of the Suide sample is the largest, and its pore structure has the strongest anisotropy, which is consistent with the characterization results of the fractal dimension. Suide sample has the largest fractal dimension, which is 2.9. This shows that the structure tensor can accurately describe the distribution of the pore structure of loess.
(5) In the prediction of soil and water characteristic curves based on pore structure, the predicted air intake values of the micro-CT Luochuan-1 sample and the SEM scanning sample are similar. This is because the resolution of the two is similar, but the Luochuan-1 The pores of the sample are smaller, and the curve is more gradual. The samples from Suide, Luochuan, and Xi'an have smaller air intake values for the SEM prediction soil and water characteristic curves. Because their average pore diameters are larger than the SEM results, the curves show a steeper trend than the SEM image curve.

[1] KUKLA G, AN Z. Loess stratigraphy in Central China [J]. Palaeogeography, Palaeoclimatology, Palaeoecology, 1989, 72:203-25.

[2] Deng J, Wang L, Zhang Z, et al. Microstructure characteristics and forming environment of late Quaternary Period loess in the Loess Plateau of China [J]. Environmental Earth Sciences, 2010, 59(8): 1807-17.

[3] Rutter N, Dend Z, Evans M, et al. Baoji-type pedostratigraphic section, Loess Plateau, north-central China [J]. Quaternary Science Reviews, 1991, 10(1): 1-22.

[4] 钱程. 冻融作用下黑方台黄土力学特性及微细观结构变化研究 [D]. 北京：中国地质大学，2018.

[5] 贾小旭. 典型黄土区土壤水分布及其对草地生态系统碳过程的影响 [D]. 杨凌: 西北农林科技大学，2014.

[6] Zhao G, Mu X, Wen Z, et al. Soil erosion, conservation, and eco-environment changes in the Loess Plateau of China[J]. Land Degradation & Development, 2013, 24(5): 499-510.

[7] 王慧娟. 黄土高塬沟壑区塬面水土流失防治措施布设研究[D]. 郑州：华北水利水电大学, 2019.

[8] 闫亚景. 结构在非饱和黄土抗剪特性中的控制机理 [D]. 北京：中国地质大学, 2013.

[9] Shi H, Shao M. Soil and water loss from the Loess Plateau in China [J]. Journal of Arid Environments, 2000, 45(1): 9-20.

[10] Rahma A E, Warrington D N, Lei T. Efficiency of wheat straw mulching in reducing soil and water losses from three typical soils of the Loess Plateau, China [J]. International Soil and Water Conservation Research, 2019, 7(4): 335-45.

[11] Luo L, Lin H, Halleck P. Quantifying Soil Structure and Preferential Flow in Intact Soil Using X-ray Computed Tomography [J]. Soil Science Society of America Journal, 2008, 72(4):1058-1069.

[12] Luo L, Lin H, Li S. Quantification of 3-D soil macropore networks in different soil types and land uses using computed tomography [J]. Journal of Hydrology, 2010, 393(1): 53-64.

[13] Perret J, Prasher S O, Kantzas A, et al. Three‐dimensional quantification of macropore networks in undisturbed soil cores[J]. Soil Science Society of America Journal, 1999, 63(6): 1530-1543.

[14] Perret J S, Prasher S O, Kacimov A R. Mass fractal dimension of soil macropores using computed tomography: from the box‐counting to the cube‐counting algorithm[J]. European Journal of Soil Science, 2003, 54(3): 569-579.

[15]Monga O, Ngom F N, Delerue J F. Representing geometric structures in 3D tomography soil images: Application to pore-space modeling[J]. Computers & geosciences, 2007, 33(9): 1140-1161.

[16] Torrance J K, Elliot T, Martin R, et al. X-ray computed tomography of frozen soil[J]. Cold regions science and technology, 2008, 53(1): 75-82.

[17] Kawaragi C, Yoneda T, Sato T, et al. Microstructure of saturated bentonites characterized by X-ray CT observations[J]. Engineering geology, 2009, 106(1-2): 51-57.

[18] Ngom N F, Monga O, Mohamed M M O, et al. 3D shape extraction segmentation and representation of soil microstructures using generalized cylinders[J]. Computers & geosciences, 2012, 39: 50-63.

[19] Beckers E, Plougonven E, Roisin C, et al. X-ray microtomography: a porosity-based thresholding method to improve soil pore network characterization?[J]. Geoderma, 2014, 219: 145-154.

[20] Ojeda-Magaña B, Quintanilla-Domínguez J, Ruelas R, et al. Identification of pore spaces in 3D CT soil images using PFCM partitional clustering[J]. Geoderma, 2014, 217: 90-101.

[21] Scheibe T D, Perkins W A, Richmond M C, et al. Pore‐scale and multiscale numerical simulation of flow and transport in a laboratory‐scale column[J]. Water Resources Research, 2015, 51(2): 1023-1035.

[22] Wieland R, Rogasik H. Method for analyzing soil structure according to the size of structural elements[J]. Computers & Geosciences, 2015, 75: 96-102.

[23] Müller K, Katuwal S, Young I, et al. Characterising and linking X-ray CT derived macroporosity parameters to infiltration in soils with contrasting structures[J]. Geoderma, 2018, 313: 82-91.

[24]Ivanov A L, Shein E V, Skvortsova E B. Tomography of soil pores: From morphological characteristics to structural–functional assessment of pore space[J]. Eurasian Soil Science, 2019, 52(1): 50-57.

[25] Wei Y, Fan W, Yu B, et al. Characterization and evolution of three-dimensional microstructure of Malan loess[J]. Catena, 2020, 192: 104585.

[26] 李晓军. 路基填土破损过程的细观识别与破损参数测定 [D]. 西安：长安大学, 1999.

[27] 冯杰, 郝振纯. CT 扫描确定土壤大孔隙分布[J]. 水科学进展, 2002, 13(5): 611-617.

[28] 张建丰, 林性粹, 王文焰. 黄土的大孔隙特征和大孔隙流研究 [J]. 水土保持学报, 2003, 17(4): 168-71.

[29] 雷胜友, 唐文栋, 王晓谋, 等.原状黄土损伤破坏过程的CT扫描分析(Ⅱ)[J].铁道科学与工程学报,2005(01):51-56.

[30] 倪万魁, 杨泓全, 王朝阳. 路基原状黄土细观结构损伤规律的 CT 检测分析[J]. 公路交通科技, 2005, 22(6): 81-83.

[31] 雷胜友, 唐文栋. 黄土在受力和湿陷过程中微结构变化的 CT 扫描分析[J]. 岩石力学与工程学报, 2004, 23(24): 4166-4169.

[32] Dong H, Blunt M J. Pore-network extraction from micro-computerized-tomography images[J]. Physical review E, 2009, 80(3): 036307.

[33] Silin D, Patzek T. Pore space morphology analysis using maximal inscribed spheres[J]. Physica A: Statistical mechanics and its applications, 2006, 371(2): 336-360.

[34] 连江波, 张爱军, 郭敏霞, 等. 反复冻融循环对黄土孔隙比及渗透性的影响[J]. 人民长江, 2010, 41(12): 55-58.

[35] 谷天峰, 王家鼎, 郭乐, 等. 基于支持向量机的Q_3黄土孔隙微观结构研究 [J]. 水文地质工程地质, 2010, 37(06): 102-6.

[36] 吕菲. 土壤孔隙结构三维重建软件系统的设计与实现 [J]. 山西建筑, 2011, 37(12): 251-3.

[37] 蒋运忠, 王云亮, 汪时机. 在 MATLAB 环境下实现膨胀土 CT 图像的三维重建 [J]. 西南大学学报: 自然科学版, 2011, 33(3): 144-8.

[38] 郑剑锋, 马巍, 赵淑萍, et al. 三轴压缩条件下基于 CT 实时监测的冻结兰州黄土细观损伤变化研究 [J]. 2011, 33(4): 839r-45.

[39] 程亚南, 刘建立, 吕菲, 等. 基于CT图像的土壤孔隙结构三维重建及水力学性质预测 [J]. 农业工程学报, 2012, 28(22): 115-22.

[40] 马骏骅, 马可, 徐贵娃.冻融循环作用下黄土状土孔隙分布分形几何研究[J].煤炭工程,2012(S2):129-132.

[41] 计璐艳, 耿修瑞, 赵永超.基于土壤显微图像时间序列的三维重建方法[J].电子测量技术,2013,36(08):61-64.

[42] 王宇, 李晓, 阙介民,李守定.基于CT图像灰度水平的孔隙率计算及应用[J].水利学报,2015,46(03):357-365.

[43] 徐宗恒, 徐则民, 李凌旭. 基于CT扫描的斜坡非饱和带土体大孔隙定量化研究和三维重建 [J]. 水土保持通报, 2015, 35(01): 133-8.

[44]Li T C, Shao M A, Jia Y H. Application of X‐ray tomography to quantify macropore characteristics of loess soil under two perennial plants[J]. European Journal of Soil Science, 2016, 67(3): 266-275.

[45] 姚志华, 陈正汉, 李加贵，等. 基于CT技术的原状黄土细观结构动态演化特征[J].农业工程学报,2017,33(13):134-142.

[46] 鞠忻倪, 贾玉华, 甘淼, 等. 黄土沟壑区不同地形部位土壤大孔隙特征研究 [J]. 土壤学报, 2018, 55(05): 1098-107.

[47] Li J, Li X. Spatial Statistics of Large Pores in Loess[C]//GeoShanghai International Conference. Singapore：Springer, 2018: 802-810.

[48] Li Y, He S, Deng X, et al. Characterization of macropore structure of Malan loess in NW China based on 3D pipe models constructed by using computed tomography technology[J]. Journal of Asian Earth Sciences, 2018, 154: 271-279.

[49] Wang J, Qin Q, Guo L, et al. Multi-fractal characteristics of three-dimensional distribution of reconstructed soil pores at opencast coal-mine dump based on high-precision CT scanning[J]. Soil and Tillage Research, 2018, 182: 144-152.

[50] Yu X, Peng G, Lu S. Characterizing Aggregate Pore Structure by X‐Ray Micro‐Computed Tomography and a Network Model[J]. Soil Science Society of America Journal, 2018, 82(4): 744-756.

[51]Yi J, Yang Y, Liu M, et al. Characterising macropores and preferential flow of mountainous forest soils with contrasting human disturbances[J]. Soil Research, 2019, 57(6): 601-614.

[52] Li X, Lu Y, Zhang X, et al. Quantification of macropores of Malan loess and the hydraulic significance on slope stability by X-ray computed tomography[J]. Environmental Earth Sciences, 2019, 78(16): 1-19.

[53] 解朝阳, 张巍, 姚东方, 等. 基于最大球算法定量表征土体空间孔隙网络 [J]. 工程地质学报, 2020, 28(1): 60-8.

[54] 刘慧. 基于CT图像处理的冻结岩石细观结构及损伤力学特性研究 [D]; 西安: 西安科技大学, 2013.

[55] 杨亚静. 基于 MC 算法的医学图像三维重建的研究 [D]. 天津：河北工业大学, 2012.

[56] Levoy M. Display of surfaces from volume data[J]. IEEE Computer graphics and Applications, 1988, 8(3): 29-37.

[57] 耿丽娟, 李星毅. 用于大数据分类的 KNN 算法研究 [J]. 计算机应用研究, 2014, 31(5): 1342-4.

[58] Wei T, Fan W, Yuan W, et al. Three-dimensional pore network characterization of loess and paleosol stratigraphy from South Jingyang Plateau, China[J]. Environmental Earth Sciences, 2019, 78(11): 1-15.

[59] Muhunthan B, Chameau J L. Void fabric tensor and ultimate state surface of soils[J]. Journal of geotechnical and geoenvironmental engineering, 1997, 123(2): 173-181.

[60] 李洁梁. 黄土大孔隙的矢量化表征与软件开发 [D]. 西安：西安科技大学, 2019.

[61] Bullock P, Thomasson A J. Rothamsted studies of soil structure II. Measurement and characterisation of macroporosity by image analysis and comparison with data from water retention measurements[J]. Journal of Soil Science, 1979, 30(3): 391-413.

[62] Tippkötter R. Morphology, spatial arrangement and origin of macropores in some hapludalfs, West Germany[J]. Geoderma, 1983, 29(4): 355-371.

[63] Vogel H J. A numerical experiment on pore size, pore connectivity, water retention, permeability, and solute transport using network models [J]. European Journal of Soil Science, 2000, 51(1): 99-105.

[64] Li P, Xie W, Pak R Y S, et al. Microstructural evolution of loess soils from the Loess Plateau of China[J]. Catena, 2019, 173: 276-288.

[65] Nawaz M F, Bourrie G, Trolard F. Soil compaction impact and modelling. A review[J]. Agronomy for sustainable development, 2013, 33(2): 291-309.

[66] Otani J, Mukunoki T, Obara Y. Application of X-ray CT method for characterization of failure in soils[J]. Soils and Foundations, 2000, 40(2): 111-118.

[67]Elliot T R, Reynolds W D, Heck R J. Use of existing pore models and X-ray computed tomography to predict saturated soil hydraulic conductivity[J]. Geoderma, 2010, 156(3-4): 133-142.

[68] Campbell P, Abhyankar S. Fractals, form, chance and dimension[J]. The Mathematical Intelligencer, 1978, 1(1): 35-37.

[69] 朱华, 姬翠翠. 分形理论及其应用 [M]. 北京：科学出版社, 2011.

[70] CHEN X, ZHOU J. Volume-based soil particle fractal relation with soil erodibility in a small watershed of purple soil [J]. Environmental earth sciences, 2013, 70(4): 1735-46.

[71] 于东明, 胡小兰, 张光灿, et al. 江子河小流域不同植被类型土壤粒径的多重分形特征 [J]. 中国水土保持科学, 2011, 9(5): 79-85.

[72] Martı́n M Á, Montero E. Laser diffraction and multifractal analysis for the characterization of dry soil volume-size distributions[J]. Soil and Tillage Research, 2002, 64(1-2): 113-123.

[73] 杨世玉, 赵人达, 靳贺松. 地聚物砂浆的力学性能与孔结构分形特征分析 [J]. 华南理工大学学报 (自然科学版), 2020, 48(3): 126.

[74] 徐永福. 非饱和膨胀土结构性强度的研究 [J]. 河海大学学报(自然科学版), 1999, (02): 89-92.

[75] 陶高梁. 岩土多孔介质孔隙结构的分形研究及其应用 [D]. 武汉：武汉理工大学, 2010.

[76] 宋维琪, 徐月森, 张云银, 等. 基于傅里叶级数分析的各向异性参数估计及裂缝预测 [J]. 石油物探, 2020, 59(1): 108-13.

[77] 陶高梁,张季如.表征孔隙及颗粒体积与尺度分布的两类岩土体分形模型[J].科学通报,2009,54(06):838-846.

[78]张季如, 陶高梁, 黄丽，等.表征孔隙孔径分布的岩土体孔隙率模型及其应用[J].科学通报,2010,55(Z2):2661-2670.

[79] [1]陶高梁, 孔令伟, 肖衡林, 等.土-水特征曲线的分形特性及其分析拟合[J].岩土力学,2014,35(09):2443-2447.

[80] Li X, Hu C, Li F, et al. Determining soil water characteristic curve of lime treated loess using multiscale structure fractal characteristic[J]. Scientific Reports, 2020, 10(1): 1-11.