International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004
Where, Æ is the multi-looking number, $, j is the phase
of pixel ^7 11.
The direct-recursive algorithm is based on the principle
that the absolute value of the difference between two adjacent
pixels is smaller than 7 . In order to describe this algorithm
easily, we suppose the phase sequence is (9. y . The
calculation is described as follows:
a) calculate the phase difference: Ag = 9, = 9, ;
b) adjust the difference to I= 4. x):
Aó-27T ^ó»-m
Ad AG +27 Ad « —t ;
Ag others
¢) calculate the unwrapped phase pixel by pixel:
$,, 74 * ^d;
In order to show the process of direct-recursive algorithm,
we suppose that the phases in an unwrapping sequence are list
as follows: 'Unit in cycle(T]
0.1110.3710.4(10.3(10.7(10.90J0.1110.2......
The difference between two adjacent pixels is respectively
list as:
0.2.10.173-0.1710.4(.0.2 1-0.8010.1......
Where, the value -0.8 should be added 1 because it’s
absolute value is larger then 0.5 the according phase is 77 [1 So,
the phase difference should be -0.8+1.0 [1 0.2. The phase
sequence is corrected to:
0.2010.101-0.110.400.2770.200.1......
The unwrapped results of the phase sequence are:
0.1110.3710.430.30J0.710.9011.111.2......
The direct-recursive phase unwrapping method is simple
and easily affected by noises, but it can be successful for the
filtered fringe in the high coherence region or high fake
coherence region grown by the seed pixel.
In order to testify the validity of the new phase unwrapping
algorithm composed by vector filtering and region grow phase
unwrapping method, this paper has done a lot of experiments
using different data. The results show that the new method can
be used successfully in the phase unwrapping processing for
different data. The detailed data and experimental results are
introduced in the next section.
4. EXPERIMENT
4.1 THE DATA OF THE EXPERIMENT
[n this experiment, the ERS-1/2 data obtained in 1996 is
used, which covers some plain area in China. Another data used
is obtained by the SIR-C in 1994, which covers some
fluctuation area of Tianshan in China. The wavelength of
ERS-1/2 is 5.66 cm, while that of SIR-C is 24 cm.
4.2 RESULTS
Some raw interferograms of ERS-1/2 and of SIR-C as well
as their filtering results by different filtering methods are shown
>
in figure6 and figure 7. The interferogram in Figure 6a is the
Internati
raw fringe of SIR-C. Figure 6b-6f show the fringes filtered by
different methods. Accordingly, the fringe derived from
ERS-1/2 data and its filtered results are shown in Figure 7a-7f.
Figure 6b interferogram Filtered by
Mean Filtering(7x7)
Eieeee 4o Does iete Tecos
Figure
by ^
Figure 6c interferogram Filtered Figure 6d Filtered interferogram m
by Median Filtering(7x7) by Adaptive Filtering Boils
has t
So, 1
vectc
fake
fring
cohe
phas:
Figure 6e interferogram Filtered Figure 6f interferogram Filtered
by Multi-looking filtering by Vector Filtering(7x7)
Figure8a r
Figure 7b interferogram
Figure 7a Raw interferogram filtered by Mean Filtering
104