Image Processing Reference
mechanisms developed to select the right checker pattern before other better solutions
Spiral and Bayer Dithering
The spiral dither approach takes a rectangle of pixels and orders the pixel drawing algorithm
so that it draws a spiral pattern as you progress from one color to another. The image is col-
ored more smoothly but has a somewhat grainy appearance. Because the grain is evenly dis-
tributed across the image, you don't get the edging artifacts that the checkerboard dithering
gives you. The Bayer dither improves on this and yields a more consistent coverage.
This describes a variety of dithering spirals and incremental fill matrices. The values
are used to determine a threshold against which to fill a pixel array. This is one of many
so-called patterned dithering techniques.
Error Diffusion Dithering
In 1976, Robert W. Floyd and Louis Steinberg published a paper called “An Adaptive
Algorithm for Spatial Greyscale.” This described a new approach to dithering that
scanned across the image and decided whether a pixel was light or dark not by simply slic-
ing through the intensity graph, but by accumulating an error value from the surrounding
pixels. It was developed to diffuse or distribute quantization errors over a larger area and
it yielded some amazing results. Even when using 1-bit black and white displays, photo-
Figure 6-12 Spiral and Bayer dithering.
Bayer dithering (open source code): http://www.cs.rit.edu/usr/local/pub/pga/Graphics/GIF/gr/dither.c