The Image Analysis On Sand Particles Characteristics From Microscopic

Sand is a typical discrete material, the discrete characteristic of microscopic determines the macroscopic anisotropy. Therefore, the research of sand fabric from the microscopic has attracted great attention in the world. There are three microscopic causes of sand fabric, namely, the preferred orientation distribution of the normal vector of particle contact, the spatial arrangement of particles and the preferred orientation distribution in the spatial of void. This article mainly aimed at the sand particle contact point vector and particle space arrangement of fabric two causes to study, aiming at some deficiencies in the existing research, redefined the theoretical expression of the two, combining the theory of image recognition and matlab software, based on the morphology of the principle’s image analysis technology, providing two causes of sand microscopic fabric the quantitative description. Using the advantages of electronic and communications disciplines in microscopic graphical analysis and combined with the geotechnical macro microscopic theory knowledge, for the scientific providing basis for granular material macro microscopic combining quantitative research.First of all, the thesis is puts forward the methodology of what causes the normal vector of sand particle contact. Based on the existing research of the fabric for the normal vector of the particle contact, it provides a general description for normal vector of particle contact in 3D space:First, quantitative description of the fabric in the 2D space is deduced by means of morphological image recognition technology; next, the invariants of the tensor in 2D can define the size and direction of anisotropy of the main fabric; then description the 3D is formed, that is grain contact information in 2D space can be employed to describe its fabric features in 3D space. With the increase of fabric order, this method can precisely describe graphical information, overcomes many deficiencies of the traditional method and widens the application range.Secondly, this thesis is offers the methodology of what causes spatial arrangement of the sand particle. On the basis of fabric tensor unable to describe the influence of the shape of sand particle has in its anisotropy, it utilizes morphological image processing technology to transform any shape of the sand particle into elliptic-shape one to definite its geometry size. Since major and minor axes of ellipse do influence particle arrangement, the general theory of fabric tensor for particle arrangement is refreshed. In addition, morphological image processing and analyzing technology contributes to the quantitative description of fabric tensor.Thirdly, the image processing analysis program and compiled to analyze tested images of sand micro-structure. The image of the processing technology is employed to process the collected SEM images. This process includes the pretreatment, binaryzation, median filtering, morphological processing and segmentation for the adhesion particles, etc., making SEM image become a clear binary one with particle separation and boundary. Matlab software also used to adjust the program with the different image processing algorithms and compare the different processing rendering algorithm statements, so that the most suitable algorithm is chosen for the particle image processing and gives further preparation for the next step of image analysis with its well processed pictures.Finally, based on morphological image recognition technology, compiled the two kinds of sand fabric image processing and analysis program and parameters of the microscopic sand fabric characteristics are extracted. To be exact, these parameters should include the particle contact normal vector, the tangent vector, and data analysis, statistics as well as simulation from the program. On the other hand, it should include the elliptic of secondary order under elliptical particle pixel sense and the major and minor axis of this one under pixel sense. Then the relevant data needs to be extracted to calculate the size of the fabric tensor. Meanwhile, knowledge of probability statistics should be applied to account the microscopic parameters of particle arrangement.