GAN Based Sample Simulation for SEM-Image Super Resolution

Maoke Yang, Guoqing Li, Chang Shu, Pan Zhao, Hua Han

Abstract: We propose to employ image super resolution to accelerate collection speed of scanning electric microscopes (SEM). This process can be done by collecting images in lower resolution, and then upscale the collected images with image super-resolution algorithms. However, because of physical factors, SEM-images collected in different resolution changed not only in their scale, but also with noise level and physical distortion. Consequently, it is hard to obtain training dataset. In order to solve this problem, we designed a generative adversarial network (GAN) to fit the noise of SEM images, and then generate realistic training samples from high resolution SEM data. Finally, a fully convolutional network have been designed to perform image super-resolution and image denoise at the same time. This pipeline works well on our SEM-image dataset.

External IDs:dblp:conf/cccv/YangLSZH17