Jacobi's solution involves three consequetive coordinate transformations :
a) Spatial transformation of object-space control into camera coordinate system (origin at the
interior perspective center O, and Z azis along the optical axis).
b) Projective transformation from camera coordinate system into an intermediate image
coordinate system (plane perpendicular to the optical axis and at a distance from point O equal
to the principal distance). Correction for radial symmetrical lens distortion is then carried out
in this plane using appropriate polynomials.
c) Projective transformation of the corrected intermediate image coordinates into the adopted
image coordinate system (x, y), by rotationsa and g around point O.
In the linearization of the error equations, Jacobi uses partial differences rather than
partial differentials, thereby reducing the computer time necessary for the adjustment. à
Image
g y
P necs s (point of symmetry)
incipal point)
(interior perspective center)
Fig.
3.3. Wolpert's Calibration Method
Woólpert [ 25] designed a portable test object consisting of a number of graduated glass
scales arranged in a horizontal straight line pattern on a wall. By using multiple exposures and
rotating the camera around its axis, a second dimension is obtained. The object distance is
accurately measured, thus enabling complete calibration of the camera.
3.4. Hallert's Grid Method
This method is well documented in the photogrammetric literature, see for instance
[14] , [15], [16] and [19] . A two-dimensional test field contains a great number of targets
arranged in a concentric circles pattern. Photography is taken from a point on the perpendicular
through the central point of the field, with the negative plane approximately parallel to the plane
of the test field. The relative location of the targets are determined to a high accuracy by geodetic
measurements. The radial distortion effects are determined from the scale variation of circular
combination of test points around the center point, adjusting each circle combination separately. #
Polynomials (usually of 5th order) can also be used for the determination of radial distortion
effects, Both methods are frequently used simultaneously for checking purposes,
To determine the camera constant, either the camera or the object must be translated
an accurately determined distance along the perpendicular to the grid plane, The resulting
variation in image scale is used to determine the camera constant,
3.5. Dóhler's Calibration "Star" e I",
An interesting device was designed by Dóhler [12] as a test object for close-range camera
calibration. It consists of a metallic staff of 2.4 m length which can rotate around a mid-point
—26--