In this contribution, a definition of percolation is given which is compared with other approaches known from the literature. Furthermore, we present a method for determining the percolation from 3d image data obtained by $\mu$CT methods. We recall the notation of connectedness in the continuous case and define connected components in data given on point lattices (grids). Our method for measuring percolation is based on an effective labeling of the pore space which uses a run length encoding of the foreground pixels as well as on an effective handling of the label conflicts occurring during processing. Finally, the edge effects resulting from the windowing of the image data must be handled carefully. We make use of the macroscopically homogeneity of the investigated porous materials and apply a specific censoring of the data known from spatial statistics.
The applicability of the presented method is shown for $\mu$CT-images of a wide range of porous materials. The computation time, the necessary memory space and the influence of the lateral resolution on the results are addressed in detail.