Gamma ISP: Reference Manual

rascc_mask_thinning

NAME
rascc_mask_thinning - Adaptive sampling reduction for phase unwrapping validity mask.

SYNOPSIS
rascc_mask_thinning <ras_in> <in_file> <width> <ras_out> [nmax] [thresh_1] ... [thresh_nmax]

<ras_in>	(input)validity mask (8-bit SUN rasterfile or BMP format image)
<in_file>	(input)file used for adaptive sampling reduction, e.g. coherence (float)
<width>	number of samples/row of in_file
<ras_out>	(output) validity mask with reduced sampling (8-bit SUN rasterfile or BMP format image)
[nmax]	number of sampling reduction runs (default: 3)
[thresh_1]	first threshhold (used for smallest scale sampling reduction)
[...]	further threshholds
[thresh_nmax]	threshhold nmax (used for largest scale sampling reduction)

EXAMPLE
rascc_mask_thinning 01608_02109.mask.ras 01608_02109.cc 2456 01608_02109.mask_thinned.ras 5 0.3 0.4 0.5 0.6 0.7

Reads the input phase unwrapping validity mask and reduces the sampling adaptively to the provided coherence. Five thinning runs are conducted.

DESCRIPTION
rascc_mask_thinning Reads the input phase unwrapping validity mask and reduces the sampling adaptively to the provided coherence. In low coherence areas (i.e. for high phase noise or high phase gradient) the full sampling is maintained. In areas of higher coherence (i.e. for lower phase noise and lower phase gradients) the sampling is reduced.

The main reason for such adaptive thinning of the phase unwrapping validity mask is to reduce the number of samples considered in the global optimization step of the Minimum Cost Flow (MCF) phase unwrapping technique. On one hand this technique is memory intensive, on the other hand gllobal optimization is wanted. In many cases masking and adaptive thinning of the validity mask allow to reduce the number of samples allowing to solve the MCF optimiation problem in a single step, i.e. without patching of the data, which is the necessary aproach otherwise.

The thinning is done in several runs. In a first run the sampling is reduced in areas where the reference file value, typically selected to be a coherence image, is above the first threshhold indicated. In a second run additional thinning is applied for areas where the reference file value is above the second threshhold indicated, and so on. Consequently, the threshold values indicated should rather increase, as larger scale thinning is only applicable for very low phase variation.

The targeted thinning is that the phase variation between the retained points is small as compared to a full phase cycle. Undersampling has to be avoided, as this will increase the difficulty of the phase unwrapping.

The input validity mask is generated using the program rascc_mask and it is used for phase unwrapping with the program mcf, to avoid regions of high phase noise in the phase unwrapping.

For all pixels corresponding to masked pixels (i.e. pixels not to be considered in the phase unwrapping) the validity mask values are set to 0, respectively to black, i.e. (0,0,0).

The program interp_ad can be used to interpolate the unwrapped phases with the reduced sampling. In order to have the full interferogram unwrapped these interpolate the unwrapped phases can be used as a model and assuming that the unwrapped phase value is always with an interval of (-PI,+PI) of the model value. This step is supported by the program unw_model and permits to achieve an unwrapped phase fully corresponding to the wrapped phase in the sense that it is identical if re-wrapped.

OPTIONS
None.

DIAGNOSTICS
All messages are generally self-explanatory.

NOTES
None.

SEE ALSO
mcf, rascc_mask, interp_ad, unw_model.