Research On Algorithm Of 3D Anisotropic Media Reconstruction Based On 2D Image

Using computer software to reconstruct 3D structure digital core has the characteristics of low price,convenient,reusable and so on.It has a very broad application prospect,and can provide valuable theoretical and technical support for production practice in many industries.In this paper,the 3D reconstruction algorithm of anisotropic media is studied,and the training image analysis method and three effective reconstruction methods of anisotropic 3D media are proposed,mainly including:(1)Morphological completeness analysis method of two-dimensional image.Through the analysis of two-dimensional training images,the specific completeness screening conditions and analysis methods are improved and supplemented.According to the number of training images in different situations,the best training image control mode can be determined according to this analysis method,so as to ensure better reconstruction effects.(2)The chord length distribution function is used to reconstruct the anisotropic threedimensional structure from a single two-dimensional image.Aiming at the reconstruction of anisotropic three-dimensional media in a single image,a simulation algorithm of chord length distribution function is proposed to reconstruct the anisotropic three-dimensional media,and the control mode of chord length in each direction is given in the two-dimensional and threedimensional reconstruction.This method mainly uses the chord length distribution function to extract the chord length features in each direction of the training image,which is a method to extract the statistical features of the training image.It can accurately ensure that the reconstructed structure and the reference structure have the same anisotropic characteristics.(3)A two-layer multilevel grid system is used to reconstruct the anisotropic threedimensional structure from a single two-dimensional image.For the reconstruction of a single image,another reconstruction idea is given,which realizes the construction method of doublelayer multilevel grid and the mode density function suitable for double-layer multilevel grid.Under the premise of guaranteeing the reconstruction speed,the ability of the template to represent the image shape is improved,and the section shape of the reconstructed structure is closer to the training image.(4)Using grid construction method to reconstruct anisotropic 3D structure from multiple2 D images.A second training image was introduced to improve the shape of each section of anisotropic three-dimensional structure,and a grid construction method was proposed to effectively solve the problem of porosity mismatch caused by controlling different sections of different training images.Then,a pattern replication method with orthogonal strings as constraints is presented to improve the reconstruction accuracy of the third section.By introducing the second training image to control the third section,this reconstruction method further improves the reconstruction effect.The comparison of the morphology of 3D structure slices reconstructed by the above three methods shows that the three methods proposed in this paper can effectively solve the reconstruction problem of anisotropic 3D media,and can maintain the similarity of the morphological characteristics of the reconstructed structure and the reference structure.Through the quantization functions such as local porosity function,two-point correlation function and linear path function,the method proposed in this paper can effectively maintain the similarity of statistical characteristics between the reconstructed structure and the reference structure.The results of visual analysis and concrete quantification show that the reconstruction method proposed in this paper is successful.