Effective Super-Resolution Method for Paired Electron Microscopic Images

This paper is concerned with developing super-resolution algorithms for handling electron microscopic images. We note two main aspects differentiating the problem discussed here from existing approaches in the super-resolution literature. The first difference is that in the electron imaging setting, we have a pair of physical high-resolution and low-resolution images, rather than a physical image with its downsampled counterpart. The high-resolution image covers about 25% of the view field of the low-resolution image, and the objective is to enhance the area of the low-resolution image where there is no high-resolution counterpart. The second difference is that the physics behind electron imaging is different from that of optical (visible light) photos. The implication is that super-resolution methods trained by optical photos are not going to be effective when applied to electron images. Focusing on the unique properties, we propose a new super-resolution image algorithm entailing the following components: (a) a global and local registration method to match the high- and low-resolution image patches; (b) a clustering method that selects the representative patches for a paired library, and (c) a non-local-mean method that reconstructs the high-resolution image using the paired library. Tested on 22 pairs of electron microscopic images, we find that the proposed method outperforms considerably the existing super-resolution methods. A noteworthy merit of the proposed method lies in its superior ability to suppress background noise without deteriorating the quality of the foreground signal.