File type: DjVu Image
Developed by Paul G. Howard, Patrick Haffner, L’on Bottou, and by Yann Le Cun in 1996 at the AT&T Laboratories, the DJVU file extension and its accompanying technology are marketed by LizardTech. T
his format file represents a compressed format for image files designed to provide content providers with a way of scanning high resolution colored pages and images without the matching large file size. In this aspect, it is a viable alternative for use in publishing catalogs, magazines, books, and similar media over the Internet.
It introduces a tremendously better compression algorithm in dealing with images by compressing colored JPEG or GIF image files five to ten times better than other standards and three to eight times better compression when dealing with black and white TIFF-based documents. Another distinct advantage of the DJVU file extension is that it is a free file format based on the Open-Source licensing concept.
The source code for the file format’s reference library can be downloaded from the host website.
The ownership for the commercial development rights for the encoding software has gone from one company to another in the recent years. The originators of this compressed file format maintain an implementation based on GPL. This is called DjVuLibre.
Category: Raster Image File
File format: Open-Source
Open with Windows: WinDjView, Lizardtech DjVu Solo, Web browser with LizardTech DjVu Browser Plug-in
Open with Linux: N/A
The DJVU file extension operates by dividing one image into varying images and then applying separate compression for each instance of the image. In creating a DJVU format-based image, the first image must be separated into various components classified into a mask image, foreground image, and background image.
Foreground and background images normally fall under the classification of low resolution images (100 dots per inch or less). While the mask image must be a high resolution, bi-level image (300 dots per inch or higher) which normally contains stored text.
The foreground and background images are compressed via wavelet-based algorithm commonly called IW44.
The mask image on the other hand undergoes a compression method named JB2 which is similar to the JBIG2 algorithm.
The JB2 encoding standard is responsible for identifying nearly all of the identical shapes contained in the page like in the case of multiple instances of specific characters based on a particular size, style, and font type.
The compression of the bitmap is then based on the uniqueness of its shape and is done separately. It proceeds by encoding the location where every shape can be seen in the page.