didjvu

DjVu encoder with foreground/background separation

The DjVu graphics format is very effective because it uses multiple layers which are differently compressed. For the derivation of the bitonal foreground layer ("mask") of (scanned document) images ("segmentation") didjvu uses the Gamera framework. In the same process didjvu encodes the foreground mask together with the background counterpart into DjVu. That's also possible for a stack of images which can be processes into a bundled DjVu container very easily. The results are suitable for getting high- quality OCR data even from problematic scans from old documents. Furthermore, didjvu is capable of processing bitonal segmentation on colour scans.