IAPRS & SIS, Vol.34, Part 7, “Resource and Environmental Monitoring", Hyderabad, India,2002
GE
8. VERIFICATION OF REGISTRATION
The complex interferogram was computed and 2:10 multi-look
averaging done. Finally, the phase-difference image was
computed and displayed (Fig. 4).
9, CONCLUSIONS
The algorithm takes a time of 70 seconds plus the time needed
to resample the slave image, to register the images which have
1250 columns and 4400 rows, on a PENTIUM III Processor. KEY
The following factors contribute to the speed-up in the
algorithm: Corresponding points are computed for points at
intervals of 100 pixels in the azimuth and range directions in ABST
the master image. The cross-correlation function needs to be
computed only for offsets in the vicinity of the possible range
; ; : ; Image
and azimuth displacements between the images. This speeds up develc
the computation. Also, it is sufficient to interpolate a single cell
of the cross-correlation matrix in order to compute the under
displacement accurate to 1/20". of a pixel. A 5 * 5 window of °P atio
the cross-correlation matrix was used for computing the B India;
coefficients of the cubic B-splines for the interpolation and a hasıor
fast algorithm was used for the interpolation. paper
: compr
REFERENCES
Hou, H.S. and Andrews, H.C., 1987. Cubic Splines for Image
Interpolation and Digital Filtering, IEEE Trans. ASSP, ASSP- LL D
26, pp. 508-517. Increa:
; . resulte
Leber, F.W., 1990. Radargrammetric Image Processing, and si
becom
small «
Artech House, Massachusetts.
Y M
THE
A
Amon;
widely
Discre
entrop:
> upon
coeffic
factor,
compre
quality
indirec
image |
result
inform
data dc
the use
desired
quantiz
approat
tables,
An att
quantiz
sensing
of errc
S nU ds pnt i j^ Ts : dr
AREA SURE : itness
A bis
Figure 4: Fringe Pattern
