Compression Algorithms

TIFF

The Tag Image File Format specification was released in 1986. by Aldus Corporation (now Adobe System Inc.). Its' main purpose is to provide storing and interchange of raster images created by scanners and desktop applications. TIFF is not a printer or page description language. If you need to format an image in BMP, GIF, JPEG, scanned fax document or any non-vector image, and want a rich description of that image with all details, then TIFF is right choice for you.

First labeled revision was the 3.0. Next version - 4.0 contained minor enhancements and was published in 1987. The 5.0 revision, released in October 1988, added support for palette color images and LZW compression. The main purpose of revision 6.0, published in June 1992, was to support CMYK and YCbCr color images, and the JPEG compression.

Features :

portability (works under MS-DOS, Unix, Mac)
scalable color depth from 1-bit (b/w) to 24-bit (true color)
numerous compression schemes :
- uncompressed
- RLE
- LZW (GIF)
- CCITT Group 3&4 Fax
- JPEG
- PackBit
max file size of 4096 MB
multiple images per file
extensibility for future

Structure

File structure contains three sections:

the Image File Header (IFH)
the Image File Directory offset (IFD)
bitmap data

The IFH consist of 8 bytes: first 2 bytes specify the byte order, second 2 specify the file version, and last 4 bytes point on the first IFD. The IFH must be at the beginning of the file, while IFD may be at any location in file.

An IFD contains an image description and pointer on actual image data. It has three sections: counter of fields (tags) in current IFD, tags and pointer on next IFD, if such exists.

Tags

Tags consist of 12-byte entry. First two bytes are field identification tag, followed by two determining bytes for the type of tag data. Next four bytes count items in the tag data and last four are tag data (if they fit in 4 bytes), or pointer on data that is located anywhere in the file. Tags must be sorted in ascending order. The IFD can contain numerous tags, therefore its' baselines (classes) are defined according to the type of image they store. Image readers may refuse to recognize tags if they are private or don't fit in the baseline. Tags can contain image description, page name, title, specific target printer, name of author, copyright, date of creation, number of images in file, application or hardware that created image, etc.

Image Data

Primary, TIFF contains bitmap data. Any other vector or text data require addition of new tags. An image is organized as a rectangular array of pixels divided into strips. Location of each strip can be anywhere in the TIFF file.

Examples

Four TIFF images with four different compress algorithms :

T - CCITT Group 4, monochrome
i - LZW compression, 8bit color depth
f - PackBit, true color, CMYK color sheme
f - JPEG, true color

Books

References