Applications of deep learning-based denoising methodologies for scanning electron microscope images

Abstract

Abstract In this paper, we use five types of deep-learning algorithms for denoising scanning electron microscope (SEM) measurement data. Denoising of SEM images is an important task since the images often suffer from noise, which can make it difficult to accurately interpret the data. We also investigate realistic SEM denoising characteristics using a variety of metrics to assess the quality of denoised images. Overall, we find that the trained generative models provide superior denoising performance and that it is crucial to objectively quantify the performance, just like in the scanning process itself. It is anticipated that the deep-learning based technique can accelerate image measurements, which can be utilized for very fast analytical investigations. We also demonstrate that the success of a generative model may depend on the appropriate assessment of noise characteristics in the specific image data analysis of interest. Moreover, it is addressed that denoising performance can be properly evaluated when a relevant metrics that aligns well with human visual systems.