Automatic fiber thickness and cloudiness analysis for non-woven filter media based on sem images

K. Schladitz*, M. Godehardt, A. Moghiseh, P. Easwaran, Fraunhofer-Institut für Techno- und Wirtschaftsmathematik ITWM, Germany

The microstructure of non-woven filter media influences the macroscopic properties of the filter significantly. Scanning electron microscopy (SEM) yields highly resolved image data of the fiber systems. However, the potential of these images is not yet fully used due to a lack of automatic image analysis solutions. Here, we present MAVIfiber2d for automatic measurement of local fiber thickness and local fiber orientation distributions based on SEM image data. To achieve these, MAVIfiber2d thrives on recent advances in fiber thickness and orientation analysis [1,2].

Moreover, MAVIfiber2d quantifies the homogeneity of the given sample with respect to orientation and coverage on several scales. To this end, two approaches are pursued – the classical one based on a pyramid decomposition [3,4] and alternatively a frequency space based approach [5]. More precisely, the images are pyramidally decomposed by Laplacians of several filter sizes. The responses on the individual scales are...

Figure 1: Synthetic SEM image of a realization of the Altendorf-Jeulin model for fiber systems from [2] and corresponding MAVIfiber2d thickness map. The true input thicknesses are 8 and 16 pixels corresponding at pixel size 1µm to 8 and 16 µm.

Session: F3 - Filter Media - Quality Control and Pore Size Analysis III
Day: 13 March 2018
Time: 16:45 - 18:00 h

Learn more at FILTECH 2018 - Register Now!