ed by string
6. DEM GENERATION
After string matching, the conjugated point pairs along
the linear features are extracted and space intersection
is applied to generate coarse DEM data. The height
information (disparity) of these linear feature points is
used as the predicted conjugated position, Based on
these coarse DEM data which offer sufficient pull-in
range, object space least squares matching can be
performed for accuracy refinement, the matching would
start at predicted conjugate position which would not
lead to mismatching, and the time of image matching
would be reduced also, thus increasing the efficiency of
fine DEM generation, and the high quality DEM data
can be obtained at the end.
The traditional method uses window of pixels for
matching to determine a single point (usually the
middle point) only, but object space least squares
matching uses a window of pixels for matching to
determine multi-points in a grid pattern DEM in one
solution [Lo,1994]. The high contrast pixels (linear
features) would offer a larger contribution to the
decision making, helping to avoid making the wrong
decision in the homogeneous part of the image. Thus,
a combination of the advantages of feature-based
matching and intensity-based matching can be obtained
for DEM generation.
7. CONCLUSION
(a) Feature-based matching is performed at feature
level rather than signal processing level which would
not be influenced by geometric distortion and
radiometric distortion if we do matching in image space
rather than object space. Therefore, by applying string
matching to extract the corresponding linear feature
pairs in conjugated epipolar line pairs is a robust
approach and the results are highly reliable.
(b) For the conventional matching strategy, the target
area of the left image is selected to search for the best
match in the search area of the right image only; the
result may be different, however, if the matching is
from right to left. String matching uses the mutual
matching strategy, which matches not only left to right
but also right to left, then the selection of the minimum
cost among them is the " really best matching".
(c) There are two ways to assess the similarity measure
by using the cost function: distance measure approach
and conditional probability approach. The distance
measure approach is applied if the attributes of
primitives are numeric. If the attribute values of
primitives are symbolic, then the conditional probability
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996
approach can be applied still. Therefore, the cost
function is a universal approach for solving
correspondence analysis problem.
REFERENCES
Boyer, K.L./Kak, A.C. [1988]: Structural Stereopsis for
3-D Vision. IEEE Transactions on Pattern Analysis
and Machine Intelligence, Vol.10, No.2. 144-166.
Heipke, C. [1992]: A Global Approach for Least
Squares Image Matching and Surface Reconstruction in
Object Space. PE&RS: Vol.58, No.3, 317-323.
Kostwinder, H.R./Mulder, N.J./Radwan, M.M. [1988]:
DEM Generation from SPOT Multiple View Images
using Correspondence Analysis. ISPRS 16th Congress,
Kyoto. Com.III, III.442-449 (B11).
Lo, K.C. [1989]: Image Quality Assessment for
Automatic Production of DEM Data. Conference of
Photogrammetry Automation, Tainan, Taiwan. J1-J20.
Lo, K.C. [1993]: Image Segmentation for Region
Matching. Journal of Surveying Engineering, Vol.35,
No.2, 63-81.
Lo, K.C. [1994]: High Quality Automatic DEM
Generation from Multiple Image. Doctor Thesis.
Twente University. The Netherlands.
Lu, S.Y. [1982]: A String to String Correlation
Algorithm for Image Skeletonization. Proc. of 6th
International Conference on Pattern Recognition,
IEEE. 178-179.
Mulder, N.J./Sijmons, K. [1984]: Image Segmentation
using Conditional Rankorder Filtering. ISPRS 15th
Congress, Rio de Janeiro. Com.III.
Wrobel, B.P. [1987]: Facet Stereo Vision (FAST
Vision) - A New Approach to Computer Stereo Vision
and to Digital Photogrammetry. Conference on Fast
Processing of Photogrammetric Data, Interlaken,
231-258.