Samples of english handwriting (characters and numbers) are essential tools when training and testing character recognition systems. While researching development of Gradematic, we had trouble finding a free handwritten character sample library, so we collected our own. These samples were collected from native English writers all over the world, in attempts to capture multiple methods of drawing numbers and letters.
There are two different sample packs available. The first sample pack, unprocessed images, are the individual characters images as they were directly extracted from a document - they have had no post-processing applied to them, meaning they may have specks, borders from the original form, and so fourth. The second sample pack has been run through our image preprocessor, part of Open ICR, in order to remove noise as well as standardized the characters to make it easier for machines to read them.
March 18, 2013: Original Release