no code implementations • 24 Oct 2019 • Dmitry Rodin, Nikita Orlov
Our method divides the document into blocks and collects luminance features from the original image and black-white strokes histograms of the binarized image.
