INTRODUCTION - Fundamentals of Three-Dimensional Digital Image Processing - page 10

extreme effort, and building up a 3D structure can be dicult. Nonetheless
the technique is an extremely important one, as it allows judgments to be
made based on direct observation of the actual sample (see Fig. 1.2).
(d) Artificial imaging
This refers to computer-based creation of the image data in their entirety,
with an array of 3D data created as a time series and used as the subject
of visualization. Some examples include the behavior of physical phenomena
based on numerical solutions to differential equations (the behavior of fluids,
deformation of materials, electromagnetic field strengths, etc.), saving image
data as voxel data for use in computer graphics, creating a virtual space and
storing it as voxel data, etc. As computer performance continues to improve,
computer simulations have rapidly become increasingly more important as
a fundamental method of scientific research. Computational physics is one
example. In such cases, it is typical for data resulting from computer ex-
periments to be created and stored as large-scale voxel structure data and
visualization tools used to make viewing of the data possible. Along with
the increasing importance of such methods, a variety of techniques has been
developed for visualization and is currently an important topic in 3D image
processing [IEEE95c, Johnson06].
(e) Range imaging
A range image is one in which a reference point (or reference plane, generally
the location of the point of view or that of the measuring device) is chosen,
and the distance from that point to the surface of all objects forward from the
point is measured and recorded. Systems measuring propagation times from
reflected lasers or ultrasound (called “range finders”) are used to find the dis-
tance to reflective surfaces. Similar devices include fish detectors, ultrasound
equipment used in medical imaging, and display screens on radar receivers.
These devices do not allow internal structures to be recorded, but can handle
applications related to external surface forms. Though these cannot be called
true 3D images, they can be thought of as 3D in the sense that they can
measure the distance to complex curved surfaces from multiple directions, the
representation of which requires the use of a 3D space. The devices have been
used in a variety of methods and areas, including the use of lasers to record the
shape of human faces and bodies, the input of simple 3D shapes, examination
of the image of human organs using ultrasound, observation of micron-order
surface features on VLSI components using confocal microscopy, etc.
A somewhat analogous scenario is to take a 2D gray-scale image f ( x, y )
as a surface in 3D space described by x, y, f . Taking this perspective one can
analyze the features of the 3D curve, which is taken as a part of 2D image
processing. Extending this to a true 3D image as discussed above makes it a
4D image (see Chapter 6).