An efficient and uniformly behaving streamline-based μCT fibre tracking algorithm using volume-wise structure tensor and signal processing techniques

Abstract

A method for reconstructing polygonal paths of fibres in reinforced composites imaged using micro-computed tomography is formally described, implemented and tested. The algorithm has been crafted to be explicable, require no training data and behave uniformly in all axes or orientations. It consists of four phases: (1) segmenting fibre regions using a scale-dependent Iterative Difference of Gaussians approach, (2) extracting directionality using the structure tensor minimum eigenvector, (3) automatically placing the seeds near a set of user-defined restricting surfaces, and (4) tracking fibres using a streamline-based integration method. The algorithm cost grows in relation to the target fibre diameter and is proportional to the number of voxels in the input volume. Its behaviour, ability to process very curved fibres, and error have been assessed using both synthetic and real datasets. The C++ implementation is performant and parallelizable, and produces helpful visualisations to gain insight of the intermediate and final results

This work has been financially supported by grants from the Spanish Government (Ministerio de Ciencia, Innovacion y Universidades) PID2019-106426RB-C31

Open Access funding provided thanks to the CRUE-CSIC agreement with Elsevier