This paper explores the use of generative adversarial networks (GANs), specifically SRGANs, to super-resolve scanning electron microscopy (SEM) images of cracked cathode materials in Li-ion batteries. The authors address the challenge of balancing image volume and resolution in characterizing fine features like cracks, which require large volumes for statistical representativity but high resolution for accurate detection. Quantitative analysis shows SRGANs outperform other networks in crack detection, making GANs a viable method for super-resolution microscopy images to enable representative quantification of fine features.

Orkun Furat, Donal P. Finegan, Zhenzhen Yang, Tom Kirstein, Kandler Smith, Volker Schmidt