Learn How Handwriter Constructs Writer Profiles
Handwriter constructs writer profiles from handwriting samples.
Handwriting Samples
Handwriter takes scanned handwriting samples saved as PNG images. For best results, scan your samples at 300 DPI.
Handwriting as Shapes
Handwriter splits the scanned handwriting into component shapes called graphs. Graphs capture shapes, not necessarily individual letters. They might be a part of a letter or contain parts of multiple letters.
Common Handwriting Shapes
Handwriter analyzed handwriting samples from 100 writers, grouping the tens of thousands of graphs into 40 forty clusters of similar shapes with a K-Means clustering algorithm. The result is a cluster template of 40 exemplar handwriting shapes. These shapes might be part of a letter or contain parts of multiple letters. The cluster template is used to estimate writer profiles.
Writer Profiles
Each graph in a handwriting sample is matched to its most similar shape in the cluster template. By counting how often a writer uses graphs from each cluster, handwriter creates a profile of a writer’s handwriting style. This profile shows the writer’s tendency to form shapes in particular ways.
