Research On 3D-Reconstruction Of Geometric Model Of Porous Materials Based On Single Image

Because of porous materials' unique pore structure and flow characteristics,they are widely used in industrial production.To a large extent,the structure of pore space determines properties of porous materials,so it is very important to analysis and research the structure of pore space.In this thesis,three-dimensional reconstruction based on the SEM images of porous material has been done.The detail study works are as follows:We obtained binary images of porous materials through pre-processing of scanned images,in which we made use of image processing technologies such as noise filtering,gray enhancing etc,therefore we can clearly distinguish the porosity of porous materials and solid framework.In this way the two important statistical parameters,porosity and self-correlation function can be easily extracted.A truncated Gaussian method based on Fourier transform is introduced to generate the 3-D porous structure.The 3-D porous structure is described by the porosity and autocorrelation function,which are obtained from the 2-D binary image. According to the theory,when the autocorrelation function is positive-definite,the corresponding Gaussian field can be generated applying Fourier transforms directly. Combining the phase angle distribution which is assumed to be random,we can get a series of Gaussian fields with the same autocorrelation and different phase angle.By making use of particle system of OpenGL,we achieved visualization of three-dimensional model.Particle system can not only be applied to simulate irregular nonintegrated structures,but also can be used in 3D-rebuilding of overall objects.We generated particles in cubic capacity by particle system and worked out their coordination through stochastic process.Making use of OpenGL,we produced the 3D geometric model of porous medium at last.Based on studying of the principles and methods of verification,we selected suitable computing software for fluid to verify the three-dimensional model established in this thesis.The results confirmed the effectiveness of the three-dimensional model based on the Fourier transform of the truncated Gaussian field method.