A skeletonization-based approach for individual fiber separation in tomography images of biocomposites

The separation individual fibers is a persistent challenge in analyzing fiber composites and fibrous materials with X-ray microtomography. A variety of approaches have been published, but they generally work poorly for heterogeneous fibers with varying cross sections, orientations, lengths and shapes. We present a skeletonization-based method that can separate highly curled and heterogeneous pulp fibers in biocomposites with thickness close to the resolution limit. Optical pulp analysis for fibers extracted from the composites is used as a reference. We show that while the mean length is underestimated by our method, the shape features are better analyzed than in the reference method as fibers are not extracted or swollen in water. Our analysis reveals that the shape factor and orientation of fibers have power law dependencies on fiber length. The fiber separation and analysis method can be used as a basis for numerical modeling of the materials.