Image Processing Reference
In-Depth Information
FIGURE 3.10: Plot of RDEV vs. n for various area estimator. (the first set).
Reprinted from the proceedings of Conference on Vision Geometry, 1832(1993), S. Chattopadhyay et al., Digital
Plane Segments, 150-161, Copyright (1993), with permission from SPIE.
line segments that are the pre-images of the given discrete straight line seg-
ment (DSLS). We also presented in detail an iterative refinement algorithm
to analyze a DSLS and compute its domain.
We also described several length estimators of digitized straight lines in
two and three dimensions. The accuracies of these estimators were also dis-
cussed. A study of the performance of the estimators shows the importance
of characterizing the chain code in a much better way. It reveals that a richer
characterization leads to better estimators, and naturally the faithful charac-
terization yields the best length estimator for 2-D and 3-D DSLS.
A netcode (like a chain code for lines) is described as a representation of
a digital plane segment. Some of its properties were examined and many area
estimators were introduced in terms of the number of different elements in the
netcode.
Exercises
1. Suppose that a point p in 3-D is denoted by (p
x
,p
y
,p
z
), and let p
z=0
denote the point (p
x
,p
y
,0).
A set of digital points in 3-D S is called even iff its projection onto the
xy-plane is one-to-one, and for every quadruple (p,q,r,s) of points in S
such that p
z=0
−s
z
) ≤ 1.
Prove that a simple digital surface is a digital plane iff it has the evenness
property.
−q
z=0
= r
z=0
−s
z=0
, we have (p
z
−q
z
)−(r
z