X-Ray Tomography Crystal Characterization: Automatic 3D Segmentation

Abstract Understanding the structural parameters of crystals during crystal growth is essential for the pharmaceutical and chemical industries. This study proposes a new method for 3D images of crystals obtained with micro X-ray computed tomography. This method aims to improve the crystal segmentation compared to the watershed methods. It is based on plane recognition at the surface of the crystals. The obtained segmentation is evaluated on a synthetic image and by considering the recognized particle number and convexity. The algorithm applied to three samples (potassium alum, chromium alum, and copper sulfate) reduced oversegmentation by 87% compared to watershed based on ultimate erosion while keeping the convexity of the recognized particle.

Gautier Hypolite, Jérôme Vicente, Philippe Moulin. X-Ray Tomography Crystal Characterization: Automatic 3D Segmentation. Microscopy and Microanalysis, 2023, 232, pp.119673. ⟨10.1093/micmic/ozad019⟩. ⟨hal-04055847⟩

Journal: Microscopy and Microanalysis

Date de publication: 16-03-2023

Auteurs:
  • Gautier Hypolite
  • Jérôme Vicente
  • Philippe Moulin

Digital object identifier (doi): http://dx.doi.org/10.1093/micmic/ozad019

x >