Image-based Data on Strain Fields of Microstructures with Porosity Defects_1

Published: 15 May 2020| Version 1 | DOI: 10.17632/s4g76zd5ys.1
Pranav Milind Khanolkar,
, Christopher McComb


This dataset is one of the three datasets that are utilized in the research, 'Predicting elastic strain fields in defective microstructures using image colorization algorithms' submitted for publication in the Computational Material Science Journal. This is a part of supplementary data for the Data in Brief article co-submitted alongside the aforementioned paper in Computational Material Science Journal. The data file is a MATLAB Binary Data Compressed file (.mat) that contains images of microstructures and their respective strain fields represented as two-dimensional arrays. This data is utilized for training a convolutional neural network to predict strain fields within a microstructure, through image colorization algorithm. This data contains image data of 1000 samples of microstructures with 100 circular porosity defects having their radii uniformly distributed in the range 0.1 – 0.5 mm, and their corresponding strain fields.



Microstructure, Strain, Deep Learning