Moving!

I have moved this blog to wordpress.com. Please, do visit blancosilva.wordpress.com for the updated version. Thanks!

Super-Resolution Micrograph Reconstruction by Nonlocal-Means Applied to High-Angle Annular Dark Field Scanning Transmission Microscopy (HAADF-STEM)

We outline a new systematic approach to extracting high-resolution information from HAADF–STEM images which will be beneficial to the characterization of beam sensitive materials. The idea is to treat several, possibly many low electron dose images with specially adapted digital image processing concepts at a minimum allowable spatial resolution. Our goal is to keep the overall cumulative electron dose as low as possible while still staying close to an acceptable level of physical resolution. We wrote a letter indicating the main conceptual imaging concepts and restoration methods that we believe are suitable for carrying out such a program and, in particular, allow one to correct special acquisition artifacts which result in blurring, aliasing, rastering distortions and noise.

Below you can find a preprint of that document and a pdf presentation about this work that I gave in the SEMS 2010 meeting, in Charleston, SC. Click on either image to download.

The Nonlocal-means Algorithm

The nonlocal-means algorithm [Buades, Coll, Morel] was designed to perform noise reduction on digital images, while preserving the main geometrical configurations, as well as finer structures, details and texture. The algorithm is consistent under the condition that one can find many samples of every image detail within the same image.

Barbara	Noise added, std=30	Denoised image, h=93

The algorithm has the following closed form: Given a finite grid $\Lambda \subset \mathbb{Z}^2$ of the form $\Lambda = \Omega \cap \mathbb{Z}^2$ for some compact set $\Omega \subset \mathbb{R}^2$, a signal $f \in L_2(\Lambda,\mathbb{R}^+)$, and a family of windows $\{ \mathcal{R}_k \}_{k \in \Lambda}$ satisfying the conditions

1. $k \in \mathcal{R}_k$ for all $k \in \Lambda$.
2. If $j \in \mathcal{R}_k$, then $k \in \mathcal{R}_j$,

the nonlocal-means operator $\operatorname{NL}_h\colon \ell_2(\Lambda,\mathbb{R}) \to \ell_2(\Lambda,\mathbb{R})$ with filtering parameter $h>0$, is defined by

\[\operatorname{NL}_h f(k) = \sum_{j \in \Lambda} \omega_h(j,k) f(j),\]

where the weights $\{ \omega_h(j,k) \}_{j,k \in \Lambda}$ are defined by

\[\omega_h(j,k) = \frac{ \exp \bigg( -\frac{\left\lVert f(\mathcal{R}_j) - f(\mathcal{R}_k) \right\rVert_{2,a}^2}{h^2} \bigg) }{ \sum_{j \in \Lambda} \exp \bigg( - \frac{\left\lVert f(\mathcal{R}_j) - f(\mathcal{R}_k) \right\rVert_{2,a}^2}{h^2} \bigg)}.\]

Here, $f(\mathcal{R})$ denotes a patch of the image $f$ supported on the window $\mathcal{R}$.

Notice that the similarity check between patches is nothing but a simple Gaussian weighted Euclidean distance, which accounts for difference of grayscales alone. Efros and Leung prove that this distance is a reliable measure for the comparison of texture patches, and at the same time copes very well with additive white noise; in particular, if $f$ and $g$ are respectively the noisy and original images, and $\sigma^2$ is the noise variance, then the most similar patches in the noisy image are also expected to be the most similar in the original:
\[ \mathbb{E} \left\lVert f(\mathcal{R}_j) - f(\mathcal{R}_k) \right\rVert_{2,a}^2 = \left\lVert g(\mathcal{R}_j) - g(\mathcal{R}_k) \right\rVert_{2,a}^2 + 2\sigma^2 .\]