Figure 3
Schematic workflow of the automatic detection algorithm. The process begins with a low-angle data subset A. Subsequently, crystallographic statistics and clustering bootstrapping are performed independently. The overlap between each subcluster Ci,j and the set O derived from crystallographic statistics is then assessed. This results in the concatenated multiset , where elements can be recurrent with a certain multiplicity. If an unique element has sufficient multiplicity, it will be included in the output set N and categorized as a NEMO. The detection performance is influenced by the hyperparameters t, l and m. Here, an element p is equivalent to indexed 2D coordinates with (x, y) as positional properties in a Euclidean plane. |