Image Processing Reference
In-Depth Information
age, by specifying appropriate values of the parameters. It is observed that
the algorithm based on the concepts of exponential averaging and straight-
ness checking produces marginally better results when the Prewitt operator
is chosen instead of the Sobel operator. For checking the robustness to image
rotation, the algorithm is tested on all the 360 images rotated from 0
0
to 359
0
,
and their outputs are recorded. It is noticed from each such output that the
sequence of straight edges describing an object in the rotated image remains
the same or nearly the same as that of the original one; occasionally, there
occurs a variation due to changes in gray-value distribution around the edges
owing to the subject rotation.
4.6 Summary
In this chapter we discuss the characterization of DSS and discuss its
application in polygonal approximation by relaxing some of the criteria for
declaring a digital arc as a DSS. It is evident from the discussion and the
algorithms that a set of ADSS extracted from a set of digital curves is signif-
icantly smaller in size than that of a DSS extracted from the same, although
each ADSS can be treated as su
ciently straight for various practical applica-
tions. The CPU time needed for ADSS extraction is remarkably less than that
for DSS extraction. For polygonal approximation, the set of ADSS serves well
to determine a suboptimal solution from an arbitrary set of digital curves. It
is observed from the experimental results and analysis that the polygon ver-
tices are densely located in and around the regions with high curvature, and
sparsely in the regions with low curvature, owing to the fact that the length
of an ADSS (alternatively, a DSS) is small in the former region but high in
the latter.
Optimizing the set of ADSS to cover a given digital curve segment is a
promising area of research, since the output set of the algorithm DETECT-
ADSS depends on the start point and the direction of the traversal. For a
gray-scale image, the straight edges can also be extracted using the properties
of digital straightness so as to preserve the geometric information of the image.
The algorithm EXTRACT-ADSS is a typical example of the application of
digital geometry for image analysis using the concept of digital straightness.
In subsequent chapters we will deal with the geometry of curved lines and
surfaces in digital space.
