Gamma LAT: Reference Manual

cc_ad

ANSI-C program: cc_ad.c

NAME
cc_ad - Adaptive coherence estimation with consideration of phase slope and texture.

SYNOPSIS
cc_ad <interf> <pwr1> <pwr2> <slope> <texture> <cc_ad> <width> [box_min] [box_max] [wgt_flag] [loff] [nl]

<interf>	(input) complex interferogram (complex coherence) filename
<pwr1>	(input) intensity image of first scene
<pwr2>	(input) intensity image of second scene
<slope>	(input) phase slope data (or -)
<texture>	(input) backscatter texture data (or - )
<cc_ad>	(output) adaptive coherence estimate file
<width>	number of samples/row
[box_min]	smallest correlation average box size (default=3.0)
[box_max]	largest correlation average box size (default=9.0)
[wgt_flag]	flag for weighting function selection (default=0) 0: constant weights 1: gausian weighting function
[loff]	offset in lines to starting line (default=0)
[nl]	number of lines to process (default=0, to end of file)

EXAMPLE
cc_ad 1352_1610.flt 1352_1610.pwr1 1352_1610.pwr2 - - 1352_1610.cc_ad 2500 3.0 9.0 0

Adaptive coherence estimation without consideration of local interferogram phase slope and local backscatter image texture. The adaptive estimation window size is between 3.0 and 9.0, with constant weights.

cc_ad 1352_1610.flt 1352_1610.pwr1 1352_1610.pwr2 1352_1610.slope 1352_1610.texture 1352_1610.cc_ad 2500 2.0 9.0 1

Adaptive coherence estimation with consideration of local interferogram phase slope and local backscatter image texture. The adaptive estimation window size is between 2.0 and 9.0, using a gaussian weighting function.

INSTALLATION
Source code cc_ad.c in ./src. For compilation adjust and use Makefile: Executable version cc_ad in ../bin

AVAILABILITY
Uses ISP type definition file typedef_ISP.h.

DESCRIPTION
cc_ad reads the complex normalized interferogram and the two corresponding backscatter intensity images and estimates for each pixel the degree of coherence.

The estimator is adaptive in the sesnse that the estimator window size is adjusted adaptively to the local degree of coherence as estimated in an initial non-adaptive coherence estimation with a fixed estimator window size.

To use an adaptive coherence estimator is motivated by the trade-off between spatial resolution of the coherence estimation and estimation bias. Large estimator windows result in lower bias but reduce the spatial resolution of the coherence estimate. Small estimator windows maintain a high spatial resolution, but lead to a high estimation bias at low coherence levels. The adaptive estimator is large when estimating low coherence values (to keep the estimation bias small) and small when estimating high coherence values.

The local phase plane is considered in the coherence estimation if the input phase slope file is provided. The phase slope file can be generated suing the program phase_slope. If no such file is provided (i.e. if - is indicated on the command line instead of a valid file name) then it is assumed that the phase is constant within the estimation window. This assumption is not correct in sloped terrain resulting in an underestimation of the degree of coherence in sloped terrain.

When providing a backscatter texture image (i.e. a valid file name is indicated on the command line and not - ) then the estimator window size is adaptive not only with respect to the initial coherence estimate but also to the backscatter texture. For high texture areas the coherence estimator window size is reduced. This adaptivity is motivated by the assumption of homogeneous targets within the estimation window.

For areas of high coherence small estimation windows (near to the lower size bx_min) are used. For the estimation of low coherence values of areas with low texture large estimation windows (close to the upper window size bx_max) are used. For the estimation of low coherence values of areas with high texture intermediate to small estimation windows are used. In this case a somewhat higher estimation bias cannot be avoided.

Constant weights or a gaussian weighting function may be selected and applied. Notice that the estimation is adaptive even with constant weights.

Tests showed that the coherence estimates optained using the adaptive coherence estimation cc_ad makes an exellent compromise between maintaining a good spatial resolution and minimizing the estimation bias optained for small estimation windows. This allows a much improved interpretation of coherence values below about 0.4.

OPTIONS
The lower and upper limits used for the estimation window sizes may be specified.

DIAGNOSTICS
All messages are generally self-explanatory.

NOTES
Noisy phase slope estimates may have a negative effect on the coherence estimation. Such problems may be detected by comparison with coherence estimates without consideration of the local phase slope.

SEE ALSO
typedef_ISP.h, phase_slope, texture.