Документ взят из кэша поисковой машины. Адрес
оригинального документа
: http://graphics.cs.msu.ru/en/node/763
Дата изменения: Sun Apr 10 00:31:50 2016 Дата индексирования: Sun Apr 10 00:31:50 2016 Кодировка: UTF-8 |
Recently 3D network models of pore spaces were successfully applied for modelling different properties of rocks. These models have become a good compromise between modelling accuracy and computational cost. But quite a little effort has been conducted in the area of 3D network extraction comparing to the variety of different modelling methods. The general process of rock's properties modelling is shown below:
In this project we take 2D scans of rock's slices and construct the network of spheres and cylinders which represent pores and throats of the pore space.
Algorithm consists of two steps:
1) Pore space segmentation
2) 3D network construction
On the first step rock scans is segmented on different materials. Then, on the second step, model of a pore space is constructed and different properties of pores and throats are collected.
The process of rock segmentation is presented below:
Some steps in 3D network extraction are presented below in 2D cartoonlike fashion. At first skeleton of the pore space is computed. Then the radius in vicinity of each skeleton vertex is determined and each vertex is classified as belonging to a pore or a throat. To approximate pores as unions of spheres the pore space is cleaned by removing throats volume. Each pore volume is approximated by spheres which are in turn approximated by a number of pyramids.
3D view of approximations (pores are drawn in their actual scale and rescaled to show throats):