INTRODUCTION - Fundamentals of Three-Dimensional Digital Image Processing - page 5

1.2.3 Cues for 3D information

Given a 2D image, or two to three 2D images (2D in the sense that they are
drawn on a plane or that the dimensionality of the media is 2), the following
types of cues contain information related to the third dimension and create a
3D effect, or a sense of depth (Fig. 1.1):
(1) In a single image
• Shading (object surface brightness or changes therein)

• Shadows (changes in brightness of an object brought about by blockage

of light upon it by another object)

• Occlusion (the hiding of a background object by a foreground object)

• Deformation of known patterns by the surface of the subject (defor-

mations of slit light due to surface irregularities, etc.)

• Drawings using perspective transformations (shapes drawn using meth-

ods of perspective)

• Blurring and contrast (for example, drawing objects far away as

blurred)

• Hues (for example, drawing objects far away with a bluish tinge)

Reflections (reflection of another object on the surface of the subject)
(2) In two or more images
•

•

A combination of multiple images of the same object, depicting a dif-
ferent angle (a multi angle image). Examples of this are pairs of images
corresponding to the information brought in by each of the right and
left eye (stereocorrespondence), or tri directional design drawings.
(3) In images that include motion
•

Objects that move quickly on the screen appear to be closer, while
those moving more slowly appear to be farther away.

Using a 2D image to understand the state of a 3D scene (the positioning
of objects, etc.) based on these limited cues is a topic of primary importance
in the field of computer vision. Depending on the cues used, these problems
have names of the form “shape from X,” for example, “shape from shading,”
“shape from texture,” etc. [Horn86] see also Section 1.4. Because these cues
contain at least some information related to the 3D object, they are sometimes
called 2.5D images.

1.2.4 Specific examples of 3D images

This section discusses several specific examples, focusing on true and near-true
3D images (see Fig. 1.2).

(a) Computed tomography (CT)

CT, an X ray-based method first made practical in the early 1970s, uses
the Radon transform to measure the strength of X-rays penetrating a human