Reconstructing granular particles from X-ray computed tomography using the TWS machine learning tool and the level set method

X-ray computed tomography (CT) has emerged as the most prevalent technique to obtain three-dimensional morphological information of granular geomaterials. A key challenge in using the X-ray CT technique is to faithfully reconstruct particle morphology based on the discretized pixel information of CT images. In this work, a novel framework based on the machine learning technique and the level set method is proposed to segment CT images and reconstruct particles of granular geomaterials. Within this framework, a feature-based machine learning technique termed Trainable Weka Segmentation is utilized for CT image segmentation, i.e., to classify material phases and to segregate particles in contact. This is a fundamentally different approach in that it predicts segmentation results based on a trained classifier model that implicitly includes image features and regression functions. Subsequently, an edge-based level set method is applied to approach an accurate characterization of the particle shape. The proposed framework is applied to reconstruct three-dimensional realistic particle shapes of the Mojave Mars Simulant. Quantitative accuracy analysis shows that the proposed framework exhibits superior performance over the conventional watershed-based method in terms of both the pixel-based classification accuracy and the particle-based segmentation accuracy. Using the reconstructed realistic particles, the particle-size distribution is obtained and validated against experiment sieve analysis. Quantitative morphology analysis is also performed, showing promising potentials of the proposed framework in characterizing granular geomaterials.

     

Fundamental Structures for the Design of Machine Vision Systems

The chicken-and-egg problem of machine vision is that (a) abstracting objects requires images to be segmented, (b) segmentation requires greyscale statistics, (c) greyscale statistics are defined by segmented regions, and this leads back to (a). New image structures called gradient runs and gradient polygons are presented. These lead to new types of three-dimensional histograms.The histogram of a whole image mixes histograms from different regions within the image. These regions and their histrograms may need separating. Global histrograms can be used for preliminary segmentations. These then generate new historagms of the statistics for more coherent locally structured regions. These ideas have used to binarise documents for automated document reading systems, and they may be developed in the desingn of machine vision systems for other areas of application.     

Porous Structure Reconstruction Using Convolutional Neural Networks

The three-dimensional high-resolution imaging of rock samples is the basis for pore-scale characterization of reservoirs. Micro X-ray computed tomography (µ-CT) is considered the most direct means of obtaining the three-dimensional inner structure of porous media without deconstruction. The micrometer resolution of µ-CT, however, limits its application in the detection of small structures such as nanochannels, which are critical for fluid transportation. An effective strategy for solving this problem is applying numerical reconstruction methods to improve the resolution of the µ-CT images. In this paper, a convolutional neural network reconstruction method is introduced to reconstruct high-resolution porous structures based on low-resolution µ-CT images and high-resolution scanning electron microscope (SEM) images. The proposed method involves four steps. First, a three-dimensional low-resolution tomographic image of a rock sample is obtained by µ-CT scanning. Next, one or more sections in the rock sample are selected for scanning by SEM to obtain high-resolution two-dimensional images. The high-resolution segmented SEM images and their corresponding low-resolution µ-CT slices are then applied to train a convolutional neural network (CNN) model. Finally, the trained CNN model is used to reconstruct the entire low-resolution three-dimensional µ-CT image. Because the SEM images are segmented and have a higher resolution than the µ-CT image, this algorithm integrates the super-resolution and segmentation processes. The input data are low-resolution µ-CT images, and the output data are high-resolution segmented porous structures. The experimental results show that the proposed method can achieve state-of-the-art performance.     

Application of high resolution X-ray computed tomography to mineral deposit origin,evaluation, and processing

High resolution X-ray computed tomography (CT) is the industrial equivalent of medical CAT scanning and provides a mechanism for non-destructive studies of the three-dimensional nature of geological materials. HRXCT produces two-dimensional images (slices) that reveal the interior of an object as if it had been sliced open along the image plane for viewing. A CT image is generated by differences in X-ray attenuation that arise from differences in density and composition within the object. By acquiring a contiguous set of slices, volumetric data for all or part of a sample can be obtained, allowing three-dimensional inspection and measurement of features of interest.CT is particularly effective in the study of metallic ores that commonly contain minerals spanning the range of densities of natural materials. Available software can produce grain shape, size, and orientation data from the scanned volume, which can be particularly useful for oriented samples. CT is particularly useful for studies of gold and other precious metal-bearing minerals that typically have significant contrast even with common metallic mineral phases. CT precisely defines the in-situ location of mineral grains of interest within a sample, which then can be studied in conventional petrographic sections, and other forms of data collected, e.g. isotope or trace element geochemistry. Another area of application is fluid inclusions, which can be discerned even in opaque phases, and fluid and vapor volumes measured for sufficiently large occurrences.We summarize CT principles and review available instrumentation, as well as scanning and data reduction protocols that provide unique three-dimensional information for diverse ore types and applications in mineral deposits geology. CT applications for economic geology and other fields will continue to expand as instruments continue to evolve and as scanning protocols and applications are extended for more precise quantification of three-dimensional relationships, particularly for fine-grained particles and small fluid inclusions in larger volumes and for separation of minerals with limited contrast.     

Transferable wavelet method for grain‐size distribution from images of sediment surfaces and thin sections,and other natural granular patterns

In images of sedimentary or granular material, or simulations of binary (two‐phase) granular media, in which the individual grains are resolved, the complete size distribution of apparent grain axes is well‐approximated by the global power spectral density function derived using a Morlet wavelet. This approach overcomes many limitations of previous automated methods for estimating the grain‐size distribution from images, all of which rely on either: identification and segmentation of individual grains; calibration and/or relatively large sample sizes. The new method presented here is tested using: (i) various types of simulations of two‐phase media with a size distribution, with and without preferred orientation; (ii) 300 sample images drawn from 46 populations of sands and gravels from around the world, displaying a wide variability in origin (biogenic and mineralogical), size, surface texture and shape; (iii) petrographic thin section samples from nine populations of sedimentary rock; (iv) high‐resolution scans of marine sediment cores; and (v) non‐sedimentary natural granular patterns including sea ice and patterned ground. The grain‐size distribution obtained is equivalent to the distribution of apparent intermediate grain diameters, grid by number style. For images containing sufficient well‐resolved grains, root mean square errors are within tens of percent for percentiles across the entire grain‐size distribution. As such, this method is the first of its type which is completely transferable, unmodified, without calibration, for both consolidated and unconsolidated sediment, isotropic and anisotropic two‐phase media, and even non‐sedimentary granular patterns. The success of the wavelet approach is due, in part, to it quantifying both spectral and spatial information from the sediment image simultaneously, something which no previously developed technique is able to do.     

基于图像描述的煤岩裂隙CT图像多尺度特征

基于煤岩孔隙系统多尺度结构特征对深入认识多尺度流体运移机制的重要性,提出了基于图像描述的煤岩CT图像孔隙结构的多尺度精细描述方法。采用了图像的多策略分割技术提取目标,利用Freeman链码对目标的边界进行表达,研究了由形态学、统计矩、链码、计盒维数构造目标之间的关系、目标占有区域与边界的图像描绘子、以及分形描绘子;综合运用上述方法对煤岩CT图像中的大尺度宏观裂纹目标、小尺度细观裂隙目标进行了识别。结果表明,宏观裂纹可由灰度阈值法实现目标提取;小尺度细观裂隙需采用较复杂的分割策略,如基于索贝尔梯度算子的分水岭变换;进一步应用链码表达、图像描绘子和分形描绘子,实现了煤岩孔隙结构在欧氏空间与分形空间的多尺度精确描述。     

Dual-energy X-ray computer axial tomography and digital radiography investigation of cores and other objects of geological interest

An X-ray multi-purpose dual-energy computer tomograph has been used to investigate a variety of objects of geological interest such as unconsolidated sediments (sand, silt, clay, mud) and/or metamorphic rock samples. By using an appropriate set of standard materials as well as a proper version of filtered back projection reconstruction algorithms, we have obtained for each studied sample two images, depicting with an error of 2 to 3% densities as well as the effective atomic number. In the case of unconsolidated sediment samples, the computer tomograph has been also used to obtain good quality radiographs, prior to obtaining the tomographic images of the cross-sectional area of interest. Both radiographs and computer axial tomography images had a spatial resolution of 0.4 to 0.5 mm. Digital radiographs as well as tomographic images of unconsolidated sediment samples show with clarity internal features such as grain sizes, laminae thickness and orientation, bioturbations, revoked materials or fragments of wood. Computer tomography analysis of rock samples reveals rheological features in the relationship with metamorphism and deformation, image contrast regarding density and atomic effective number of constituents, as well as significant correlations between physical parameters and internal structures or textures.     

Experimental investigation on the grain-scale compression behavior of loose wet granular material

The behavior of model granular materials (glass beads) wetted by a small quantity of liquid forming capillary bridges is studied by one-dimensional compression test combined with X-ray computed tomography (XRCT) observation. Special attention is paid to obtain very loose initial states (initial void ratio of about 2.30) stabilized by capillary cohesion. XRCT-based analyses involve spherical particle detection adapted to relatively low-resolution images, which enable heterogeneities to be visualized and microstructural information to be collected. This study on an ideal material provides an insight into the macroscopic compression behavior of wet granular materials based on the microstructural change, such as pore distance distribution, coordination number of contacts, coordination number of neighbors and number of contacts per grain.

     

Quantifying fracture geometry with X-ray tomography: Technique of Iterative Local Thresholding (TILT) for 3D image segmentation

This paper presents a new method—the Technique of Iterative Local Thresholding (TILT)—for processing 3D X-ray computed tomography (xCT) images for visualization and quantification of rock fractures. The TILT method includes the following advancements. First, custom masks are generated by a fracture-dilation procedure, which significantly amplifies the fracture signal on the intensity histogram used for local thresholding. Second, TILT is particularly well suited for fracture characterization in granular rocks because the multi-scale Hessian fracture (MHF) filter has been incorporated to distinguish fractures from pores in the rock matrix. Third, TILT wraps the thresholding and fracture isolation steps in an optimized iterative routine for binary segmentation, minimizing human intervention and enabling automated processing of large 3D datasets. As an illustrative example, we applied TILT to 3D xCT images of reacted and unreacted fractured limestone cores. Other segmentation methods were also applied to provide insights regarding variability in image processing. The results show that TILT significantly enhanced separability of grayscale intensities, outperformed the other methods in automation, and was successful in isolating fractures from the porous rock matrix. Because the other methods are more likely to misclassify fracture edges as void and/or have limited capacity in distinguishing fractures from pores, those methods estimated larger fracture volumes (up to 80 %), surface areas (up to 60 %), and roughness (up to a factor of 2). These differences in fracture geometry would lead to significant disparities in hydraulic permeability predictions, as determined by 2D flow simulations.     

Soil porosity resulting from the assemblage of silt grains with a clay phase: New perspectives related to utilization of X-ray synchrotron computed microtomography

Understanding the assemblage of basic grains is essential for predicting many soil properties. X-ray synchrotron microtomography with a high resolution was used for visualisation and quantification of the assemblage of silt grains with a clay phase containing 20% of clays. The 3D computed image was analysed after appropriate segmentation by using: (i) autocorrelation functions which enables anisotropy discussion of the assemblage; (ii) chord distribution which gives information on the size distribution of the void and solid phases; and (iii) the Hoshen-Kopelman algorithm (6-connectivity) in order to label each void with a unique number and thus enabling discussion of the connectivity. Our results show that X-ray synchrotron microtomography enables a more accurate analysis of the assemblage of silt grains with a clay phase than with conventional methods. Thus, it showed the isotropy of the silt-clay assemblage in the studied soil material. The void and solid phase volumes were smaller than 80 and 120 μm in all directions, respectively. The mean distance between two interfaces was 6.9 and 13.4 μm for the void and solid phase, respectively. X-ray synchrotron microtomography of the soil material studied also showed that the voids resulting from the assemblage of the silt grains with the clay phase can be considered as fully connected in the studied dried materials.     

Digital measurement of 2D and 3D cracks in sandstones through improved pseudo color image enhancement and 3D reconstruction method

An integrated approach of improved pseudo color image enhancement (IPCE) method and three-dimensional (3D) reconstruction technology is proposed. The evolution characteristics of two-dimensional (2D) and 3D spatial cracks in X-ray computed tomography (CT) images of sandstones subjected to triaxial compression with confining pressure of 10, 20, and 30 MPa are investigated. The equations of crack width, length, and dip angle are established based on the proposed digital damage ratio. Based on the pseudo color images, point cloud maps with roughness and spatial images of cracks are investigated. The numerical results show that this proposed method is effective to study and understand the fracturing properties of rocks.     

Vermiculate artefacts in image analysis of granular materials

Some reported analyses of images of deforming granular materials have generated surprising vermiculate strain features which are difficult to reconcile with the mechanics of deformation of granular matter. Detailed investigation using synthetic images and improved processing of images of laboratory experiments indicates that such features can emerge as a consequence of the image acquisition (sensor, contrast, resolution), the subsequent image correlation implementation, and the user’s choice of processing parameters. The two principal factors are: (i) the texture and resolution of the images and (ii) the algorithm used to achieve sub-pixel displacement resolution. Analysis of the images using a sub-pixel interpolation algorithm that is more robust than that used originally eliminates the vermiculate features for images with moderate resolution and texture. However, erroneous features persist in images with low resolution and poor texture. Guidance is provided on ways in which such artefacts can be avoided through improved experimental and image analysis techniques.     

A new algorithm using image colour system transformation for rock grain segmentation

Summary The present paper is concerned with the development and application of an effective automatic algorithm of image analysis in order to detect grain boundaries on microscope images of Redziny dolostone and Wisniowka quartzite. The algorithm utilises sets of 6 colour images for each measurement field on thin sections, which are recorded using an optical polarizing microscope in different polarization set-ups. The proposed method is based on an image pre-processing procedure that is focused on colour system transformation, followed by rock grain boundaries segmentation using the image analysis methods for each colour image. For the image pre-processing procedure, several colour system transformations were selected and compared. By using the alternative colour systems that concentrate on colour information we are able to minimise the effects of internal micro-structures in the grain boundaries segmentation procedure. The grain boundary maps obtained confirm that the use of an approximately perceptually uniform colour system as an image pre-processing procedure can significantly improve the rock grain segmentation. This newly-developed method may facilitate petrographical and stereological studies of rock structures.     

Micromechanically inspired investigation of cemented granular materials: part II— from experiments to modelling and back

Acta Geotechnica - This paper presents an experimental and analytical/numerical study of the mechanics of cemented granular materials (CGMs). This study incorporates both in situ X-ray tomography...     

基于CT无损扫描技术的冻土内部物质研究现状与分析

CT无损动态扫描技术在土壤领域的成功应用,为冻土内部结构及变化的定量化研究提供了有效手段。通过总结和分析现有关于冻土的基于CT断层扫描图像的研究成果,阐明CT扫描技术在冻土研究中的4个关键环节,即冻土样本的制备、冻土CT图像的获取、冻土CT图像处理及分析,并指出在现有CT技术条件下冻土研究中存在的问题,针对性提出了将先进的图像处理技术与高效的图像分析手段和冻土研究相结合的解决方案,为冻土内部结构的定量研究与分析提供技术指导,进而为CT无损动态扫描技术在冻土领域的广泛应用奠定基础。     

Classify-normalize-classify

Changing atmospheric conditions often result in a data distribution shift in remote sensing images for different dates and locations making it difficult to discriminate between various classes of interest. To alleviate this data shift issue, we introduce a novel supervised classification framework, called Classify-Normalize-Classify (CNC). The proposed scheme uses a two classifier approach where the first classifier performs a rough segmentation of the class of interest (COI) in the input image. Then, the median signal of the estimated COI regions is subtracted from all image pixels values to normalize them. Finally, the second classifier is applied to the normalized image to produce the refined COI segmentation. The proposed methodology was tested to detect deforestation using bitemporal Landsat 8 OLI images over the Amazon rainforest. The CNC framework compared favorably to benchmark masks of the PRODES program and state-of-the-art classifiers run on surface reflectance images provided by USGS.     

Investigation of the effect of specific interfacial area on strength of unsaturated granular materials by X-ray tomography

This paper studies the effect of interfacial areas (air–water interfaces and solid–water interfaces) on material strength of unsaturated granular materials. High-resolution X-ray computed tomography technique is employed to measure the interfacial areas in wet glass bead samples. The scanned 3D images are trinarized into three phases and meshed into representative volume elements (RVEs). An appropriate RVE size is selected to represent adequate local information. Due to the local heterogeneity of the material, the discretized RVEs of the scanned samples actually cover a very large range of degree of saturation and porosity. The data of RVEs present the relationship between the specific interfacial areas and degree of saturation and gives boundaries where the interfacial area of a whole sample should fall in. In parallel, suction-controlled direct shear tests have been carried out on glass beads and the material strength has been corroborated with two effective stress definitions related to the specific air–water interfacial areas and fraction of wetted solid surface, respectively. The comparisons show that the specific air–water interfacial area reaches the peak at about 25% of saturation and contributes significantly to the material strength (up to 60% of the total capillary strength). The wetted solid surface obtained from X-ray CT is also used to estimate Bishop’s coefficient χ based on the second type of effective stress definition, which shows a good agreement with the measured value. This work emphasizes the importance to include interface terms in effective stress formulations of unsaturated soils. It also suggests that the X-ray CT technique and RVE-based multiscale analysis are very valuable in the studies of multiphase geomaterials.

     

X-ray micro-computed tomography study of the propagation of cracks in shale during uniaxial compression

An understanding of crack propagation is critical for the development of rock mechanic models. To study the propagation of internal cracks in situ and determine their formation mechanism, a series of uniaxial compression tests on shale specimens were conducted using a novel setup that combines X-ray micro-computed tomography (X-ray micro-CT) with a uniaxial loading apparatus, which allows CT scans to be performed during compression. Macro- and micro-scale internal cracks were extracted from CT images collected after various stages of deformation through image thresholding segmentation, providing a record of the evolution of damage within the specimens, characterized by crack closure, generation, growth, and penetration. In addition, macroscopic cracks with two distinct orientations were observed and their formation mechanism was further determined. Furthermore, test results show that the distribution of pyrite grains influences the formation of cracks at the meso- and macro-scales. These results are significant for understanding crack propagation and the failure of shale.     

基于数字图像分割法的跨孔雷达走时层析成像

跨孔雷达走时层析成像主要利用雷达波的走时进行反演,走时提取的正确与否将直接影响到层析成像的效果。数字图像分割法基于凸集投影（POCS）方法,使用能量比彩色图像分割技术准确提取走时。数字图像分割法提取走时首先应用在折射地震波的数据处理中。笔者首次将数字图像分割法提取走时的方法应用到跨孔雷达走时层析成像中,使用迭代线性反演算法重建了雷达波速度场。反演过程中,使用最小二乘QR分解法（LSQR）求解线性方程组,利用弯曲射线追踪技术构建雅可比矩阵,走时的计算值则由多模板快速推进算法（MSFM）得到。为了验证数字图像分割法在走时层析成像中的效果,使用一组合成数据及一组实测数据分别对比了基于数字图像分割法和基于传统能量比法获得的层析成像反演结果。对比结果表明,使用数字图像分割法得到的层析成像结果更为精确,误差更小,能为判断地下雷达波速度场提供更为有力的帮助。