Page 184 - Intelligent Communication Systems

CHAPTER 12/COMPUTER VISION 155

FIGURE 12.6 Dither matrix.

The binary dither method is applied to pixels with a gray-level value of 0 to a half,
and 0 or a half is assigned to the value of the pixel.
In the same way, the method is applied to pixels with a gray-level value of a
half to 1, and a half or 1 is assigned to the value of a pixel. The four-value dither
method is applied to the pixel with a value of 0 to one-third, one-third to two-thirds,
and two-thirds to 1.

12.2.8 Color Compression
We shall present one of the easiest color compression algorithms. Consider an
image with 8-bit gray level for r, g, and b. The region of color, which exists in RGB
space created along orthogonal axes r, g, and b, is divided into 256 regions as fol-
lows. An image is processed four times. From the first time to the third, the region
is divided. The fourth time, a color is assigned to the pixel.
The compression process steps are performed in the following sequence.
(1) The maximum value and the minimum value of r are determined, and the
difference between the two values is divided into eight regions.
(2) The difference between the maximum and minimum values of g is divided
into eight regions. Then the region of r and § is divided into 64 regions.
(3) The difference between the maximum and minimum values of b is
divided into eight regions. Then the region of r, g, and b is divided into
256 regions.
(4) A decision is made as to which region of r, g, and b each pixel belongs to,
and the pixel is represented by the (r, g, b) value of the selected region.
(5) Each region is represented by the mean value of r, g, and b in the region.

12.3 IMAGE TRANSFORMATION

Mathematical methods for two-dimensional image transformation are described in
this section. By one method, digital image transformation is performed via a space
filtering operation. In another method, the filtering operation is performed by con-
version from the space domain to the spectrum domain.