Fig. 5: Synthetic microstructure reconstruction utilizing a conditional generative cGAN model and DDPM. | Communications Materials

Fig. 5: Synthetic microstructure reconstruction utilizing a conditional generative cGAN model and DDPM.

From: Analyzing microstructure relationships in porous copper using a multi-method machine learning-based approach

Fig. 5

a, d and g illustrate the segmented microstructures for the porous materials HPA, HPB, and NPC obtained with FIB-SEM. b, e and h correspond to the predicted (synthetic) microstructures utilizing the cGAN model for HPA, HPB, and NPC, respectively. The data visualized in c, f and i correspond to the predicted (synthetic) microstructures utilizing the DDPM for HPA, HPB, and NPC, respectively. The synthetic and experimentally retrieved (real) microstructures are plotted for different sinter temperatures. The segmented copper and pore phases are illustrated in white and black, respectively. The frame colors are related to the porous materials (HPA, HPB and NPC), real and predicted microstructures.

Back to article page