Segmentation of turbulent computational fluid dynamics simulations with unsupervised ensemble learning

作者：

Highlights：

• Computer vision tools can be used to automate the analysis of physics simulations.

• Region of interest boundaries from image segmentation results are often fluctuating.

• Multiple clustering results can be combined with mask stacking operations.

• Ensemble machine learning methods increase stability of clustering and segmentation.

摘要

•Computer vision tools can be used to automate the analysis of physics simulations.•Region of interest boundaries from image segmentation results are often fluctuating.•Multiple clustering results can be combined with mask stacking operations.•Ensemble machine learning methods increase stability of clustering and segmentation.

论文关键词：Plasma,Turbulence,Clustering analysis,Image segmentation,Unsupervised machine learning

论文评审过程：Received 6 November 2020, Revised 30 April 2021, Accepted 23 August 2021, Available online 30 August 2021, Version of Record 9 September 2021.

论文官网地址：https://doi.org/10.1016/j.image.2021.116450