Mathematical Programming techniques for Scanning Transmission Electron Microscopy Image Segmentations

  1. J. Calvino
  2. M. López Haro
  3. J. M. Muñoz Ocaña
  4. J. Puerto
  5. A. M. Rodríguez Chía
Proceedings:
XXXIX Congreso Nacional de Estadística e Investigación Operativa; XIII Jornadas de Estadística Publica (SEIO 2022)

Publisher: Universidad de Granada

ISBN: 978-84-09-41628-8

Year of publication: 2022

Pages: 118

Type: Conference paper

GOOGLE SCHOLAR

Abstract

Electron tomography is a technique for imaging three-dimensional structures of materials at nanometer scale. Nano-objects are reconstructed using projections provided by an electron microscope from different tilt angles in order to identify the elements that constitute the object under study. This recognition procedure is known as segmentation which consists in classifying the image intensities into different clusters.The main idea of this work is to apply the discrete ordered median problem in the fiel of electron tomography to classify the pixels that constitute the images. We show specific formulations taking advantage of the problem structure depending on the image characteristics. These formulations provide good results in terms of segmentation quality and computing time.