Space Intersection by Collinearity
Space Intersection by Collinearity
If space resection is used to determine the elements of exterior orientation for both photos of a
stereopair, then object point coordinates for points that lie in the stereo overlap area can be calculated.
The procedure is known as space intersection, so called because corresponding rays to the same object
point from the two photos must intersect at the point, as shown in Fig. 1. To calculate the coordinates of
point A by space intersection, collinearity equations of the linearized form can be written for each new
point, such as point A of Fig. 1. Note, however, that since the six elements of exterior orientation are
known, the only remaining unknowns in these equations are dXA, dYA, and dZA. These are corrections to
be applied to initial approximations for object space coordinates XA, YA, and ZA, respectively, for ground
point A.
The linearized forms of the space intersection equations for point A are
(1)
(2)
Fig.1 Space intersection with a stereopair of aerial photos.
Two equations of this form can be written for point a1 of the left photo, and two more for point a2 of the
right photo; hence four equations result, and the three unknowns dXA, dYA, and dZA can be computed in a
least squares solution. These corrections are added to the initial approximations to obtain revised values
for XA, YA, and ZA. The solution is then repeated until the magnitudes of the corrections become
negligible.
Again because the equations have been linearized using Taylor’s theorem, initial approximations
are required for each point whose object space coordinates are to be computed. For these calculations,
with normal aerial photography vertical photos can be assumed, and the initial approximations can be
determined by using the parallax equations. Note that because the X, Y, and Z coordinates for both
exposure stations are known, for making these computations H can be taken as
The coordinates that result from parallax equations are in the arbitrary system described in (Let these
coordinates be designated as x′ and y′.) To convert them to the X and Y ground system, 2D conformal
coordinate transformation can be used. For this transformation, the two exposure stations can serve as
the control because their X and Y coordinates are known in the ground system, and their x′
and y′ coordinates in the parallax system are , and . Since there are four equations
and four unknowns, the transformation parameters can be solved for directly and applied to the
imaged points to get the horizontal coordinate initial approximations.
Solved Example
A stereopair of images taken with a 152.057-mm-focal-length camera has exterior orientation
parameters shown in the first table below. The images of four points have measured photo coordinates
shown in the second table below. Use the analytical space intersection to find the ground
coordinates of each of the points.
Solution
The first line is the focal length, the second two lines contain the exterior orientation
parameters of the photos, and the remaining lines contain the point name and x and y
coordinates of the point on each photo.
The object space, or ground coordinates and their corresponding standard deviations are in
meters. The photo coordinate residuals are in millimeters. The x photo coordinate residuals are
nearly equal to zero for both the left and right photos. This is expected for a stereopair
displaced in the x direction because x parallax determines the Z coordinate in object space. The
y photo coordinate residuals indicate the presence of y parallax, and their sizes are good
indicators of the quality of the solution. Generally, y coordinate residuals less than about 5 or 6
micrometers (μm) indicate a rather good solution.