Logo MAVIfiber2d
© Fraunhofer ITWM
Logo MAVIfiber2d

MAVIfiber2d is a software for geometric characterization of fiber systems based on SEM or optical microscopic images. MAVIfiber2d is particularly useful for the automatic measurement of fiber thickness based on SEM images.

MAVIfiber2d – Measuring fiber density, fiber orientation, and Cloudiness

The grade of nonwoven fabrics depends on the distribution of the fiber density, fiber orientation, and the cloudiness. These properties are evaluated in the laboratory using image data. MAVIfiber2d meets the difficult challenge of automating this evaluation, while also ensuring that it is reproducible.


MAVIfiber2d offers

  • pre-processing: smoothing, contrast enhancement
  • automatic measurement of fiber thickness in each fiber pixel without segmentation of individual fibers
  • result: area or length weighted thickness distribution, export as csv file
  • visualization: input image, result of pre-processing, thickness map
Fibersystem, Simulated SEM image, Measured thickness, Measured orientations
© Fraunhofer ITWM

Additionally can be offered

Fiber orientation analysis:

  • Measurement of local fiber orientation in each fiber pixel without segmentation of individual fibers
  • Result: area or length weighted orientation distribution, degree of anisotropy, preferred direction, export as a csv-File
  • Visualization: orientation map


Cloudiness analysis:

  • Cloudiness is a weighted mean over 7-8 scales. The user can chose between preset weights or specifying them, as the weight of the results on the different scales depends highly on the specific application.
  • Result: cloudiness at all chosen scales, cumulated cloudiness, export as a csv file
  • Visualization: cloudiness map for one scale.

Objective, Reproducible Evaluations of Nonwoven Samples

Diffusion filters have been the nucleus of image processing at ITWM and the VQC project, which measured the cloudiness of nonwovens and was one of the first industry projects in the image processing department. MAVIfiber2d combines this experience with new tools of mathematical morphology and the typical point concept of stochastic geometry to create software for objective, reproducible evaluations of nonwoven samples.


Local Analysis without Fiber Separation

In a random closed set, the typical point concept makes it possible to measure the density and distribution of fiber orientation without having to separate the fibers in the image. In unclear situations, it is not necessary to decide where each intersecting or looped fiber begins or ends. Rather a simple binarization suffices: for every image pixel, a decision is made as to where it belongs in the fiber system, i.e., the foreground or the background. Local density and orientation are defined for every foreground pixel. The result is an area-weighted distribution of density and orientation.


Cloudiness based on Normalized Grayscale Variances

Cloudiness is mathematically not that clearly defined. MAVIfiber2d builds on the VQC project findings. The input image is smoothed stepwise by approximated Gauss filters. The grayscale variances of the normalized filtered images reflect the cloudiness for the scale under consideration. The cloudiness index is calculated from the variances as a weighted average. The scales and weights are chosen such that measuring results meet as well as possible both the technical requirements and the subjective visual impression.


  • Easwaran, P.; Lehmann, M. J.; Wirjadi, O.; Prill, T.; Didas S.; Redenbach, C.:
    Automatic Fiber Thickness Measurement in SEM Images Validated Using Synthetic Data.
    Chemical Eng. & Techn. 39 (3), 395-402, 2016.
  • Altendorf, H.; Didas, S.; Batt, T.:
    Automatische Bestimmung von Faserradienverteilungen.
    In F. Puente Leon, M. Heizmann (Eds.): Forum Bildverarbeitung, Karlsruhe: KIT Scientific Publishing, 2010.
  • Altendorf, H. ; Jeulin, D.:
    3d Directional Morphology for Analysis of Fibre Orientations.
    Image Analysis & Stereology 28 (3), 2009.