83

For quick applications using normal and wide angle cameras,

the baseline to (the nearest) object distance ratio should be

roughly 2:10.

Further topics for research in basic stereoscopy are :

tF the generation of a lost stereo mate :

If the object is still accessible, the missing stereo mate

preferable should be taken from the real object. If at least a

model of the object is available, the single picture can be

projected onto the model and the stereomate can be taken.

Further research is necessary to solve this task by computer.

• the geometric potential of (available) stereo imagery ;

considering the effect of the so called stereoscopic

exaggeration :

A first idea of solving this problem can be obtained from

dealing with stereo imagery, based on normal case

photography, taken with the unknown baselength b, and an

unknown focal length c. In case of a measured object depth

Ahl and the corresponding x-parallax Apxl, such imagery

allows to calculate Ahi from Apxi for different features

according to the parallax formula

Ahi = F x Apxi

with F^Ahl / Apxi = (h/c) x (h/b)

Of course, strongly this formula is limited to differential model

areas.

A devinition of an assumed object depth Ahl has a constant

influence on the scale factor F. It influences the scale of the

object depths Ahi as calculated from horizontal parallaxes Apxi

in the same manner.

Additional topics for research in basic stereoscopy beside others

are :

• the complete history of stereo imagery, including the

stereoplasticon;

• the benefit of combining day-and night stereomates;

• the benefit of stereo panorama imagery;

• the effect of different pixelsizes for the stereomates;

• the presentation of terrestrial stereoimage blocks;

• the extention of the optical sharpness within a model, if

different focal lengths for the stereomates have been

applied;

• the simulation of the perception of the “natural”

appearance of the object dimensions;

• the market potential for stereo imagery;

3.3 Basic Education in Imagerectification

Of similar importance for undergraduate education in

Fotomatics is the rectification of single images of plane objects,

based on 4 controlpoints. This 2 dimensional plane to plane

transformation, nowadays digitally based on PC, can be

achieved independent of the focal length.

Videoclips, clearly indicating the target, successfully have been

used, to replace sketches and signals for controlpoints for

rectification and triangulation purposes !.

3.4. Basic Education in Image-Triangulation

However, the need to consider the focal length for 3dimensional

photogrammetric purposes is not enough understood in practice.

The general idea of this requirement for the focal length for 3D

Photogrammetry can be devired from a forward intersection

sketch of at least 2 images . In a photogrammetric normal case

situation the triangle to determine the object point P is defined

by the given base line and the 2 adjacent angles a and p, see

Fig. 1. According to the sketch, the triangle O’O” P is not

defined, if the focal length c is unknown, because the pure

knowledge of the image coordinates x’ and x” for the image

points P’ and P” in both images is not sufficient, to define the 2

adjacent angles a and P; i.e., as if a conventional space

intersection lacks the adjacent angles to be measured with a

theodolite.

This is the reason, why snapshots of hobby Photogrammetrists,

without recording the focal length in general hinder a 3D

evaluation. This statement is also valid for more sophisticated

photogrammetric situations.

Looking on the importance of the focal length, it is difficult to

understand, why only a view Photogrammetric cameras

available show the focal length used on the image frame.

Of course, if a sufficient amount of object coordinates is known,

the unknown focal length subsequenltty can be calculated.

image 1 image II

Fig. 1: Need of focal length for forward intersection purposes

(plane view)

For practical applications there is no need to know the base

length a priori. In the photogrammetric normal case the base

length can be derived only from 2 given object points shown in

both images.

In addition the image triangulation task can be visualized from

multiple forward intersection, with unknown base lengths.

Consequently the relation between these base lengths can be

derived from multiple point determination.

The golden photogrammetric rule for object documentation is to

verify complete object coverage with known focal length by

★ stereo synoptic photography for overview and possible

triangulation purposes,

★ single imagery for triangulation purposes led by the idea of

multiple forward intersection and

★ stereo detail photography for interpretation and geometric

purposes.

★ in addition object features, in particular object coordinates,

should be determined for geometric purposes, if possible.

Finally the forward intersection has been replaced by the bundle

block adjustment, including multiple imagewise backward

intersections, based upon collinearity equations.

OUTLOOK

Today digital image processing systems and GIS systems stay

within the good tradition of Photogrammetric instrumens. There

is no doubt, these modern tecnology in conjunction with GIS