Machine learning based de-noising of electron back scatter patterns of various crystallographic metallic materials fabricated using laser directed energy deposition

K V Mani Krishna; R Madhavan; Mangesh V Pantawane; Rajarshi Banerjee; Narendra B Dahotre

doi:10.1016/j.ultramic.2023.113703

Machine learning based de-noising of electron back scatter patterns of various crystallographic metallic materials fabricated using laser directed energy deposition

Ultramicroscopy. 2023 May:247:113703. doi: 10.1016/j.ultramic.2023.113703. Epub 2023 Feb 19.

Authors

K V Mani Krishna¹, R Madhavan², Mangesh V Pantawane², Rajarshi Banerjee², Narendra B Dahotre³

Affiliations

¹ Center for Agile and Adaptive Additive Manufacturing, University of North Texas, 3940 N Elm St, Denton, TX 76207, USA; Mechanical Metallurgy Division, Bhabha Atomic Research Centre, Mumbai, 400 085, India.
² Center for Agile and Adaptive Additive Manufacturing, University of North Texas, 3940 N Elm St, Denton, TX 76207, USA; Department of Materials Science and Engineering, University of North Texas, Denton, TX 76207, USA.
³ Center for Agile and Adaptive Additive Manufacturing, University of North Texas, 3940 N Elm St, Denton, TX 76207, USA; Department of Materials Science and Engineering, University of North Texas, Denton, TX 76207, USA. Electronic address: narendra.dahotre@unt.edu.

PMID: 36827947
DOI: 10.1016/j.ultramic.2023.113703

Abstract

A novel machine learning (ML) method of refining noisy Electron Back Scatter Patterns (EBSP) is proposed. For this, conditional generative adversarial networks (c-GAN) have been employed. The problem of de-noising the EBSPs was formulated as an image translation task conditioned on the input images to get refined/denoised output of EBSPs which can be indexed using conventional Hough transform based indexing algorithms. The ML model was trained using 10,000 EBSPs acquired under different settings for additively manufactured FCC, BCC and HCP alloy samples ensuring enough diversity and complexity in training data set. Pairs of noisy and corresponding optimal EBSPs were acquired by suitable tweaking of the EBSP acquisition parameters such as beam defocus, pattern binning and EBSD camera exposure duration. The trained model has brought out significant improvement in EBSD indexing success rate on test data, accompanied by betterment of indexing accuracy, quantified through 'pattern fit'. Complete automation of the EBSP refinement was demonstrated where in entire EBSD scan data can be fed to the model to get the refined EBSPs from which high quality EBSD data can be obtained.

Keywords: EBSD; Electron microscopy; Image processing; Machine learning.