perm filename PAGE22[00,BGB] blob
sn#046266 filedate 1973-06-05 generic text, type T, neo UTF8
~F8THE ALGORITHM: INTRODUCTION.
CRE consists of five steps: thresholding, contouring,
nesting, smoothing and comparing. Thresholding, contouring and
smoothing perform conversions between two different kinds of images.
Nesting and contouring compute topological relationships within a
given image representation. In summary the major operations are:
MAJOR OPERATION. OPERAND. RESULT.
1. THRESHOLDING: 6-BIT-IMAGE, 1-BIT-IMAGES.
2. CONTOURING: 1-BIT-IMAGES, VIC-IMAGE.
3. NESTING: VIC-IMAGE, NESTED-VIC-IMAGE.
4. SMOOTHING: VIC-IMAGE, ARC-IMAGE.
5. COMPARING: IMAGE & FILM, FILM.
Although the natural order of operations is sequential from
image thresholding to image comparing; in order to keep memory size
down, the first four steps are applied one intensity level at a time
from the darkest cut to the lightest cut (only nesting depends on
this sequential cut order); and comparing is applied to whole
images.
The illustrations on pages 21 and 23 show an initial video
image and its final smoothed contour image; the illustrations
immediately below and on page 24 the corresponding intermediate
sawtoothed contour images. The illustrated images are each composed
of seven intensity levels, and took 16 seconds and 13 seconds to
compute respectively (on a PDP-10, 2usec memory). The final CRE
data structures contained 680 and 293 nodes respectives, which comes
to 2K and 4.5K words respectively; the initial video image requires
10.2K words.
FIGURE: PUMP SAW TOOTHED CONTOURS.
~I1973,800;F8- 22 -