Aztec code

Aztec Code is a type of 2D barcode invented by Andrew Longacre, Jr. and Robert Hussey in 1995. The code was published by AIM, Inc. in 1997. Although the Aztec code was patented, that patent was officially made public domain. The aztec code is also published as ISO/IEC 24778:2008 standard. Named after the resemblance of the central finder pattern to an Aztec pyramid, Aztec code has the potential to use less space than other matrix barcodes because it does not require a surrounding blank "quiet zone".

The symbol is built on a square grid with a bulls-eye pattern at its centre for locating the code. Data is encoded in concentric square rings around the bulls-eye pattern. The central bulls-eye is 9×9 or 13×13 pixels, and one row of pixels around that encodes basic coding parameters, producing a "core" of 11×11 or 15×15 squares. Data is added in "layers", each one containing two rings of pixels, giving total sizes of 15×15, 19×19, 23×23, etc.

The corners of the core include orientation marks, allowing the code to be read if rotated or reflected. Decoding begins at the corner with three black pixels, and proceeds clockwise to the corners with two, one, and zero black pixels. The variable pixels in the central core encode the size, so it is not necessary to mark the boundary of the code with a blank "quiet zone", although some bar code readers require one.

The core of the compact Aztec code (red ascending diagonal hatching), showing the central bulls-eye, the four orientation marks (blue diagonal cross-hatching), and space for 28 bits (7 bits per side) of coding information (green horizontal hatching). The first ring of data begins outside that (grey descending diagonal hatching).

The core of the full Aztec code. 40 bits are available between the orientation marks for encoding parameters.

Message data is placed in a spiral pattern around the core. The mode message begins "01011100", indicating 01₂+1 = 2 layers, and 011100₂+1 = 29 data codewords (of 6 bits each).

The compact Aztec code core may be surrounded by 1 to 4 layers, producing symbols from 15×15 (room for 13 digits or 12 letters) through 27×27. There is additionally a special 11×11 "rune" that encodes one byte of information. The full core supports up to 32 layers, 151×151 pixels, which can encode 3832 digits, 3067 letters, or 1914 bytes of data.

Whatever part of the symbol is not used for the basic data is used for Reed–Solomon error correction, and the split is completely configurable, between limits of 1 data word, and 3 check words. The recommended number of check words is 23% of symbol capacity plus 3 codewords.

...
Wikipedia