Unsupervised microscopy image analysis and Total Variation denoising

Acquisition rates of microscopy images have reached a point where the analysis of individual images by humans, albeit computer-assisted, becomes a new bottleneck in the advancement of surface science. In this talk, I present viable alternatives to accelerate the processing of these data, including cluster analysis, pattern matching and total variation minimization. In absence of large databases, these methods constitutes a convenient alternative to data-hungry Supervised Machine Learning techniques and slow manual analysis of individual images.