4.3.2 Least Squares Matching with Geometric
Constraints: The area based process employed for the
refined matching of the point position relies on the correlation
of grey-scale patches in target and conjugate search images.
The initial estimate of corresponding pixels in conjugate
images, as provided by the epipolar matching, is required to
within one or two pixels (the "pull-in range") in order to
achieve a solution. In order to obtain an optimum accuracy
for the final point position, the area-based least squares
algorithm is supported by affine patch shaping and geometric
constraints in the form of the collinearity equations (Gruen,
1988).
4.4 Space Intersection
In addition to providing geometric constraints to the least
squares matching, the collinearity equations also serve to
produce the three dimensional (XYZ) object space co-
ordinates of the points of interest, which are being matched
after selection of the best candidate. The image matching
routine produces both digital elevation model (DEM) data in
the form of XYZ co-ordinates and error estimates for the
object space co-ordinates as a consequence of the least
squares estimation model.
During the tests carried out for the project the desired sub-
millimetre accuracy was achieved for all points in depth,
while X and Y ordinates proved typically better by
approximately a factor of ten.
4.5 Deformation Analysis
The output of the digital photogrammetric process is a dense
point cloud of three dimensional (XYZ) co-ordinates which
represent the rock face. This data is processed further in finite
element models for rock mechanics studies of the blasting
process during pre-conditioning. Results of this aspect of the
project will be reported elsewhere.
S. CONCLUDING REMARKS
The final analysis of the data made it apparent that digital
photogrammetry is highly suitable for the determination of
rock surface digital elevation models for subsequent
deformations analysis. The relative speed of the process and
the convenient size of the equipment made the technology
especially suited to the demanding underground mining
environment. Difficulties which arose from the size, shape
and environment of the object in the highly restrictive
environment, were overcome by careful planning and pre-
analysis which led to the incorporation of additional tie points
placed on target poles in front of the object. The feasibility of
this approach was confirmed by the high accuracy of the final
surface point co-ordinates.
For future application of the technique, a second reference
frame at the opposite extreme of the wall is proposed. The
digital form of the images also makes it possible to employ
image processing techniques for fracture pattern analysis and
particle analysis.
733
6. ACKNOWLEDGEMENTS
Without the assistance of Mining Technology, the rock
engineering division of the Council for Scientific and
Industrial Research (CSIR), Johannesburg, South Africa this
research would not have been possible. Thanks are extended
to them for their contribution.
For his valuable assistance in the early stages of the project
and for the use of his bundle adjustment software, the
contribution of Graeme van der Vlugt is acknowledged.
The author's are also grateful to the Foundation for Research
Development (FRD) and the University of Cape Town for
their financial support.
REFERENCES
Canny, J., 1986. A Computational Approach to Edge
Detection. IEEE Transactions on Pattern Analysis and
Machine Intelligence, Vol. PAMI-8, No. 6, pp 670-698.
Gruen, A.W. and Baltsavias, E.P., 1988. Geometrically
Constrained Multiphoto Matching. Photogrammetric
Engineering and Remote Sensing, Vol. 54, No. 5, pp 633-
641.
Haralick, R.M. and Shapiro, L.G., 1992. Computer and
Robot Vision. Vol. 1, Addison-Wesley Publishing Company,
pp 337-346.
Lim, J.S., 1984. Image Enhancement in Digital Image
Processing Techniques. Edited by Estrom, M.P., Vol. 2 in
COMPUTATIONAL TECHNIQUES edited by Alder, B.J.
and Fernbach, S., Academic Press, pp 1-51.
van der Vlugt, G. and Rüther, H., 1994. The Development of
an Automated Surface Measurement System. Paper
presented at the ISPRS Commission V, Working Group 3,
Melbourne, Australia.
Wong, K.M. and Ho, W., 1986. Close-Range Mapping with
a Solid State Camera. Photogrammetric Engineering and
Remote Sensing, Vol. 52, No. 1, pp 67-74.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996