Skip to main content
Fig. 4 | Advanced Structural and Chemical Imaging

Fig. 4

From: Feature extraction via similarity search: application to atom finding and denoising in electron and scanning probe microscopy imaging

Fig. 4

k-Means clustering applied to SVD results to identify trends in the dataset that can be invisible to the human eye. K-means clustering is applied to a subset of the U matrix obtained from SVD on the windowed dataset. a–d Images denoised using the windowing and SVD-based denoising algorithm. e–h Dendrograms showing the grouping of different clusters. The distance between two clusters is given by the height of the horizontal line connecting two clusters. i–l Maps of cluster labels obtained from K-means clustering. These maps show phase boundaries, cracks in the crystal structure, grain boundaries, and locations of contaminants in the image

Back to article page