Porous media modeling from tubular shapes and computed tomography images

This paper develops a framework for complex shape modeling with geometrical primitives. A skeleton based approximation method called Adaptive Circular Generalized cylinders (ACGC) is proposed for the description of natural porous media cavities using nondestructive images. The cavities skeleton are first approximated by spline curve running through a set of balls centers. Then, the ACGCs surfaces are constructed as rotational sweep of profile curves around the skeleton. The profile curves are sinusoidal curve envelopes solution of a partial differential equations system. The sinusoidal curve is an interpolation of the balls center and a set of contour points. An algorithm for the ACGC surface computation is proposed and computed tomography (CT) images of real soil sample was used to illustrate the proposed framework. The results lead to the conclusion that, compared to previous methods, the proposed approximation procedure gives better description of porous media cavities with more realistic shapes. Thus in future work, the ACGC representation will be used to investigate biological process inside natural porous media such as soil burrow systems and tubular organs of human body.