Evaluation of multimodal data of hydrogen distribution and crystal structures in steel sample

Abstract

The higher resolution images of the hydrogen distribution on the surface of thin stainless steel measured by electron stimulated desorption (ESD), the scanning electron microscope image, and the electron backscatter diffraction images were fused by imregister of Matlab (Mathworks). The fused data set was analyzed by the principal component analysis (PCA) and autoencoder (AE). PCA results of higher resolution data analysis showed consistent findings with the previous studies regarding the relation of crystal structure and hydrogen distribution in steel, and AE indicated more detailed information on the structures.