Little is known about metallic clusters consisting merely of a dozen of atoms or even less, despite of their importance in catalysis and crystal nucleation. Scanning transmission electron microscopy (STEM) provides direct atomic structure information but has inherently suffered from limited time resolution. We employ fast dynamic STEM combined with a spatio‐temporal image denoising algorithm to explore the structure and stability of Pt clusters on carbon, which represents a highly relevant catalysis system. At room temperature, dynamic amorphous 2D structures are found, while above ≈300 °C, the clusters transform into a crystalline state. Our experimental and theoretical data reveal an unexpected strong trend of the crystalline clusters to exhibit the face‐centered cubic, bulk structure of Pt with cuboidal geometries being most prominent.

Inverse problems in image processing, phase imaging, and computer vision often share the same structure of mapping input image(s) to output image(s) but are usually solved by different applicationspecific algorithms. Deep convolutional neural networks have shown great potential for highly variable tasks across many image-based domains, but sometimes can be challenging to train due to their internal non-linearity. We propose a novel, fast-converging neural network architecture capable of solving generic image(s)-to-image(s) inverse problems relevant to a diverse set of domains. We show this approach is useful in recovering wavefronts from direct intensity measurements, imaging objects from diffusely reflected images, and denoising scanning transmission electron microscopy images, just by using different training datasets. These successful applications demonstrate the proposed network to be an ideal candidate solving general inverse problems falling into the category of image(s)-toimage(s) translation.

We propose an effective deep learning model to denoise scanning transmission electron microscopy (STEM) image series, named Noise2Atom, to map images from a source domain S to a target domain C, where S is for our noisy experimental dataset, and C is for the desired clear atomic images. Noise2Atom uses two external networks to apply additional constraints from the domain knowledge. This model requires no signal prior, no noise model estimation, and no paired training images. The only assumption is that the inputs are acquired with identical experimental configurations. To evaluate the restoration performance of our model, as it is impossible to obtain ground truth for our experimental dataset, we propose consecutive structural similarity (CSS) for image quality assessment, based on the fact that the structures remain much the same as the previous frame(s) within small scan intervals. We demonstrate the superiority of our model by providing evaluation in terms of CSS and visual quality on different experimental datasets.

We propose an effective deep learning model to denoise scanning transmission electron microscopy (STEM) image series, named Noise2Atom, to map images from a source domain $\mathcal {S}$ S to a target domain $\mathcal {C}$ C , where $\mathcal {S}$ S is for our noisy experimental dataset, and $\mathcal {C}$ C is for the desired clear atomic images. Noise2Atom uses two external networks to apply additional constraints from the domain knowledge. This model requires no signal prior, no noise model estimation, and no paired training images. The only assumption is that the inputs are acquired with identical experimental configurations. To evaluate the restoration performance of our model, as it is impossible to obtain ground truth for our experimental dataset, we propose consecutive structural similarity (CSS) for image quality assessment, based on the fact that the structures remain much the same as the previous frame(s) within small scan intervals. We demonstrate the superiority of our model by providing evaluation in terms of CSS and visual quality on different experimental datasets.