DjVu (pronounced "déjà vu") a set of compression technologies, a file format, and a software platform for the delivery over the Web of digital documents, scanned documents, and high resolution images.
DjVu documents download and display extremely quickly, and look exactly the same on all platforms with no compatibility problems due to fonts, colors, etc. DjVu can be seen as a superior alternative to PDF and PostScript for digital documents, to TIFF (and PDF) for scanned bitonal documents, to JPEG and JPEG2000 for photographs and pictures, and to GIF for large palettized images. DjVu is the only Web format that is practical for distributing high-resolution scanned documents in color. No other format comes close.
Typical DjVu file sizes are as follows:
- bitonal scanned documents: 5 to 30KB per page at 300dpi (3 to 10 times smaller than PDF or TIFF)
- color scanned documents: 30 to 100KB per page at 300dpi (5 to 10 times smaller than JPEG)
- photos: 2 times smaller than JPEG, about the same as JPEG-2000, but the
decoder/renderer is progressive and has minimal memory requirements
- palettized images: 2 times smaller than GIF (up to 10 times if it's mostly text)
- digital (non scanned) documents: between 1 and 3 times smaller than PDF
or gzipped PS (depending on the amount of pictures), but rendering,
page flipping, zooming, panning are incomparably faster, and the image
quality on screen desplays is much better (antialiased text, etc)
More importantly, all DjVu images render very quickly and can be smoothly zoomed and panned. Pages of a document can be turned instantly, with no annoying delay. DjVu is used by hundreds of academic, commercial, governmental, and non-commercial web sites around the world to distribute scanned documents, digital documents, and high-resolution photos. DjVu was originally developped at AT&T Labs-Research. In March 2000, AT&T sold DjVu to LizardTech Inc. who now distributes Windows/CE/Mac plug-ins, and commercial encoders.
In short, DjVu is a multipage document format that can use a number of different coder/decoders (codecs) to compress the individual chunks that compose an images or a page. In fact, DjVu is really four compression techniques wrapped into one format:
- DjVuPhoto (aka IW44): A progressive, wavelet-based lossy compression
format for continuous-tone images (i.e. photos and pictures).
- DjVuBitonal (aka JB2): A lossless or lossy compression technique for
bitonal (black & white) or palettized images that is particularly
effective on images with repeated shapes (such as documents images
where the same character appears many times in the document).
- DjVuDocument: A technique for scanned color document that separates
images into a foreground layer that contains the text and line
drawings, and a background layer that contains the pictures and
background textures. The foreground is encoded with DjVuBitonal and the
Background with DjVuPhoto.
- BZZ: A general-purpose data compression technique similar to bzip2. Bzz
is used to compress searchable text layers and other metadata in DjVu
documents.
DjVu can be seen as nicely complementing PNG and MNG (which, unlike DjVu are lossless formats) in the areas of document image compression and lossy photo compression. DjVuPhoto is a bit older, but similar in spirit to JPEG-2000. It is on a par with JPEG-2000 in terms of image quality and file size, but it compares favorably in terms of rendering time and memory requirements. DjVuBitonal is better than MMR/GroupIV (used by PDF, TIFF, and most fax machines) by about a factor of 3 to 10. It is also better than the emerging standard JBIG2 by about 20%
Download the source code and learn it http://djvu.sourceforge.net