Deformation Analysis of Soil Microstructure Using Computer-assisted Tracking Technique
Wang Wei1, Hu Xin2, Hong Baoning2, He Xiaoyuan3
1College of Architecture Engineering, Nanjing, Institute of Technology,
2Research Institute of Geotechnical Engineering, Hohai University,
3Department of Engineering Mechanics, Southeast University
Abstract:Deformation properties of soil microstructure have more significant effects on many engineering characteristics. In order to reveal the movement properties of soil particles under dynamic loading conditions, computer-assisted tracking technique was presented and applied for the dynamic deformation measurement of soil microstructure. First, soil cylindrical specimen was symmetrically cut into two parts. One of the parts was vertically placed in the experimental apparatus with the glass window, which was adjacent to the plane section. Then, a long work distance telecentric microscope lens was used to monitor the movement of aggregate composed of soil particles. Using variable template correlation method, the displacements of the tracked aggregate and its locations could be obtained by identifying the maximum of correlation coefficients of each window throughout series images taken of the experiment. Finally, the test results show that the aggregate can be successfully tracked. At the same time, the deformation properties of soil microstructure under dynamic loading conditions have been effectually analyzed.
Keywords:soil microstructure; deformation properties; computer-assisted tracking technique; image correlation
1 INTRODUCTION
The development of computer technologies, coupled with a consistent improvement in quality of image forming, has led to rapid development in digital image processing and analysis. Furthermore, the powerful image processing software and low cost computer vision hardware have provided a new method for the research of civil engineering materials [1,2]. However, it is only recently that this technique has begun to be implemented in soil microstructure research [3, 4]. It is interesting to point out that the microstructure plays an important part in determining the properties and behavior of soils. Based on this, the computer-assisted tracking technique is presented in this study, which provides faster and more accurate analysis for studying deformation properties of soil microstructure.
Deformation properties of soil microstructure, such as porosity distribution, orientability evaluation,rotundity degree and fractal dimension, which are some important characteristics for researching engineering phenomena. Based on the theory of fractals, Moore and Donaldson [5] quantified soil porosity and void distribution using digital photography. Bai and Smart [6] used Scan Electron Microscope (SEM) to study the change in microstructure of Kaolin during consolidation and undrained shear. In this experiment, in order to reveal the properties of soil particles under the conditions of dynamic loading, thecomputer-assisted tracking system was used. The test results show the proposed method is an effective way to track the locations of movable aggregate. Based on the tracked images, deformation properties of soil microstructure are more accurately analyzed in tracked region than in searched one through adjusting the CCD camera by hands.
2 MEASUREMENT PRINCIPLES
2.1 Computer-assisted tracking system. The computer-assisted tracking system consists of the hardware equipment used for images capture, transformation, and storage, and the software operations such as pattern recognition, images match, etc. One of the important parts of hardware equipment is the long work distance telecentric microscope lens, with which the images photographed do not have malformation. The better quality images captured are benefit for the tracking of soil particles and the parameters extraction.
2.2 Tracking principle. The movement of soil particles or aggregate has continuity and two serial images have correlativity. Using the features, as is just mentioned, the image correlation method [7]is applied for the soil aggregate tracking. According to the gray level, the difference between two successive images is very small. The correlation coefficient can be calculated as follows:
(1)
Where is the correlation coefficient, is the greyvalue of the point in the former image, respectively, is the greyvalue of the point in the latter image. The point in the latter frame is the new position of the point in the former frame after moving the distance. The sketch of the tracking method is showed by the figure 1.
Firstly, selecting a characteristic region in the former image is taken as the template, which is used to search the movable objective, which is the characteristic region in the latter image. Then, the correspondingregion of the minimal of all the coefficients is the objective region and also regarded as thenew template instead of the former, which is utilized to search the objective in next frame. Second, according to the searching method, the objective in serial images can be tracked successively.
2.3 Deformation properties of soil microstructure. As soil is natural material, the mechanics properties of soil are very complicated. For this purpose of quantifying mechanics properties; three main parameters are introduced from microstructure to analyze soil deformation. They are porosity distribution, orientability evaluation, rotundity degree and fractal dimension. Fig. 2 shows a binary image of expansive soil. The map is amplified 200 multiple and captured by the CCD camera with microscope lens.
2.3.1 Porosity distribution. In the binary image, assuming black regions are pore area, while white ones are soil particles, then, porosity in the whole regions can be characterized by following equation (2):
(2)
Where is porosity, is pore area, and is the whole area of the image.
2.3.2Orientability evaluation. In this study, orientability evaluation of particles is applied for indicating the anisotropy of soil particles. Considering the calculation of orientability R, we also take soil aggregates as “particles” in broad sense in the binary image. So the value of orientability R can be acquired as following. It is apparent that the more is R; the better is orientability from the equation (3).
(3)
Where and are separately derived from following functions:
, (4)
Where is the angle between long axis of particle and horizontal plane.
2.3.3 Rotundity degree Rotundity degree is usually to describe the shape of soil particle in the binary image. From the following equation (5), is limited in the range of [0, 1]. If is close to 1, it will be shown that the framework of soil particle is approximately round in shape.
(5)
Where is rotundity degree. is the area of soil particle in the image. is the perimeter of soil particle in the image.
2.3.4 Fractal dimension. Fractal dimension affects the complexity of soil microstructure. Based on the fractal theory, higher fractal dimension means more complex of soil microstructure. Obviously, using the following equation (6), we can calculate the fractal dimension D.
(6)
Where is the sum of perimeters of soil particles, and is the area total of particles in the binary image.
3. EXPERIMENT
3.1 Soil sample preparation. First, cylindrical specimen of expansive clay was prepared with 80mm height and 39.1mm diameter. Second, the cylindrical sample was symmetrically cut into two parts by hand in case of destroying its natural structure. One of the parts was vertically placed in the experimental apparatus with the glass window, which was adjacent to the plane section. Meanwhile, the lateral pressure cabin was brimming with nitrogen for simulating the effect of surrounding particles. The plane sections under the loadings with different grades were photographed using a CCD camera at a resolution of pixels under the condition of surrounding stress (50KPa) and draining off water in the experiment, just as conventional triaxial test.
3.2 Experimental method. A CCD camera captures the initial image, before dynamic loading is applied on the soil specimen. Fig. 3 shows the initial soil map. The small rectangle framed with white line shown in the image is selected as initial template, which is used to search the same regions in next frame. Due to the correlation of two neighboring images, the region, which is basically same to the template, can be found in the latter image. Then, the computer-assisted tracking system drives electrical machinery to track the region and captures the image. In this case, the consecutive images photographed by the CCD camera should be the same scope researched, in which some important information can be extracted and quantified. In Fig. 4, three images tracked are shown under the states of three vertical strains. The obtained images certify to the fact that the proposed technique is a practicable technology for capturing the images of soil microstructure.
4. RESULTS AND DISCUSSION
Using the computer-assisted tracking system, we have captured a series of images about expansive soil microstructure. From the images, deformation properties of expansive soil under dynamic loading conditions can be achieved by image processing technique subsequently see Fig. 5, 6,7,8.
In figure 5, it implies that the whole tendency of porosity descends along with vertical strain increasing. It is meant that the deformation of expansive soil is very large. In fact, the results presented in the figure show the maximum of porosity can exceed 50%, which proved that expansive soil possesses abundant crack. The figure 6 clearly shows that the relation between the orientability and vertical strain. As strains are increased, the orientability of soil particles also gradually rises. Close examination of the data reveals that higher orientability will improve the properties of dilatation and contraction of expansive soil, which result in reducing of shear-strength.
It means that the soil will take place shear failure. The figure 7 is obviously proved that the rotundity degree of expansive soil is low in the early stages of loading, while high in the end stages. By the experiment, the fact is also proved that the compressibility of expansive soil is high before loading and gradually become lower with the increasing loading. Usually, fractal dimension is used to explain the complexity of soil microstructure. It is well known that the scattered structure, such as expansive soil [8] shows higher fractal dimension and has the properties of expansion with immersing in water as well as shrinkage with dehydrating. In figure 8, along with vertical strains are increased, the fractal dimension is decreasing. It means that the scattered degree of expansive soil is reduced under the dynamic loading.
5. CONCLUSIONS
Based on the image correlation method, the serial images of soil microstructure can be captured under the dynamic loading by the proposed technique, which is proved by the experiment that it is a practicable approach. Moreover, the porosity distribution, orientability evaluation rotundity degree and fractal dimension are achieved using image processing technology based on the captured images. Additionally, the deformation properties with vertical strain are also analyzed subsequently. It is recognized that the measurement precision of each parameter will be enhanced accordingly if the algorithm of computer tracking and image processing are improved.
REFERENCES
[1] Yue, Zhong Qi; Morin, and Isabelle Digital image processing for aggregate orientation in asphalt concrete mixtures. Journal of Civil Engineering, Vol.23(2), (1996), pp 480-489.
[2] Coster, M. and Chermant, J.-L. Image analysis and mathematical morphology for civil engineering materials. Cement and Concrete Composites, Vol. 23(2-3), (2001), pp 133-151.
[3] Samer Gamal Sadek. Soil structure interaction in transparent synthetic soils using digital image correlation. PhD dissertation, Polytechnic University, 2002
[4] Scott Allan Raschke. Computer vision for experimental soil micromechanics and soil characterization. PhD dissertation, The University of Michigan, 1996
[5] C. A. Moore & C. F. Donaldson. Quantifying soil microstructure using fractals. Journal of Geotechnique, Vol. 45(1), (1995), pp. 105-116.
[6] Bai, X. and P. Smart. Change in microstructure of Kaolin in consolidation and undrained shear. Journal of Geotechnique, Vol. 47(5), (1997), pp. 1009-1017.
[7] X. Y He, X. Kang, C. Quan,C. J. Tay, S. H. Wang and H. M. Shang. Optical methods for the measurement of MEMS materials and structures. Proceedings of SPIE, Vol.4537, (2002), pp 63-68..
[8] A. S. Al-Homoud, A. A. Basma, A. I. Husein Malkawi and A. Al Bashabsheh. Cyclic swelling behavior of clays. Journal of Geotechnical Engineering, Vol. 121(7), (1995), pp. 562-565.
1