4.7 Article

A tutorial on the segmentation of metallographic images: Taxonomy, new MetalDAM dataset, deep learning-based ensemble model, experimental analysis and challenges

Journal

INFORMATION FUSION
Volume 78, Issue -, Pages 232-253

Publisher

ELSEVIER
DOI: 10.1016/j.inffus.2021.09.018

Keywords

Image segmentation; Metallography; Machine learning; Deep learning; Microscopy images; Computer vision

Funding

  1. ArcelorMittal, Luxembourg Global RD
  2. ArcelorMittal Global R&D Digital Portfolio in collaboration with the Andalusian Research Institute in Data Science and Computational Intelligence (DaSCI), University of Granada
  3. Andalusian Excellence, Spain project [P18-FR-4961, SOMM17/6110/UGR]
  4. Spanish Ministry of Universities, Spain under the FPU program [FPU17/04069]
  5. Universidad de Granada/CBUA

Ask authors/readers for more resources

This paper reviews and categorizes computer vision techniques for metallographic image segmentation, introduces deep learning-based ensemble techniques utilizing pixel similarity, and conducts thorough comparisons in real-world datasets to discuss strengths, weaknesses, and application frameworks. The paper also addresses open challenges in the field to provide guidance for future research to fill existing gaps.
Image segmentation is an important issue in many industrial processes, with high potential to enhance the manufacturing process derived from raw material imaging. For example, metal phases contained in microstructures yield information on the physical properties of the steel. Existing prior literature has been devoted to develop specific computer vision techniques able to tackle a single problem involving a particular type of metallographic image. However, the field lacks a comprehensive tutorial on the different types of techniques, methodologies, their generalizations and the algorithms that can be applied in each scenario. This paper aims to fill this gap. First, the typologies of computer vision techniques to perform the segmentation of metallographic images are reviewed and categorized in a taxonomy. Second, the potential utilization of pixel similarity is discussed by introducing novel deep learning-based ensemble techniques that exploit this information. Third, a thorough comparison of the reviewed techniques is carried out in two openly available real-world datasets, one of them being a newly published dataset directly provided by ArcelorMittal, which opens up the discussion on the strengths and weaknesses of each technique and the appropriate application framework for each one. Finally, the open challenges in the topic are discussed, aiming to provide guidance in future research to cover the existing gaps.

Authors

I am an author on this paper
Click your name to claim this paper and add it to your profile.

Reviews

Primary Rating

4.7
Not enough ratings

Secondary Ratings

Novelty
-
Significance
-
Scientific rigor
-
Rate this paper

Recommended

No Data Available
No Data Available