Gamma ISP: Reference Manual

create_offset

ANSI-C program: create_offset.c

NAME
create_offset - Create or update ISP offset and interferogram parameter file

SYNOPSIS
create_offset <SLC1_par> <SLC2_par> <OFF_par> [offset_algorithm] [rlks] [azlks] [iflag]


EXAMPLE
create_offset 199600421.slc.par 19960422.slc.par 19960421_19960422.off 1 1 1 1

Use the intensity cross-correlation algorithm to measure the offsets between SLCs 19960521.slc and 19960422.slc. Enter parameters in interactive mode.

create_offset 199600421.slc.par 19960422.slc.par 19960421_19960422.off - - - 1
Use the intensity cross-correlation algorithm to measure the offsets between SLCs 19960521.slc and 19960422.slc. Enter parameters in non-interactive mode. Select default values for algorithm, rlks, and azlks parameters.

DESCRIPTION
create_offset reads the SLC parameter files and queries the user for parameters required to calculate the offsets using either the cross correlation of intensity (CCI) or fringe visibility (FV) algorithms. 

The CCI algorithm is implemented in the program offset_pwr while the FV algorithm is implements by the program offset_SLC. The SLC images are detected and a patch of selectable dimensions is correlated with a patch in the second image. Correlation of the intensity images generally is less accurate than the FV algorithm, but has the advantage that images that have no interferometric correlation, can still be co-registered. The FV algorithm extracts patches from the two SLC images and forms a small interferogram. This interferogram is Fourier transformed and the peak of the fringe spectrum is extracted. This algorithm is slower but quite sensitive even for low fringe visibilities. 

The CCI algorithm can work a larger range of offsets in the image since the offsets are measured using patches. The  patch size should be selected to be a power of 2 (32, 64, 128...)  because FFTs are used to perform the correlation matching. The FV algorithm interpolates the data by a factor of two and scans a region in half pixel offsets in range and azimuth. The accuracy of the FV and CCI approaches is about equal and CCI is almost exclusively used in practice because it is faster and works even in regions of low interferometric coherence.

Required parameters to run the co-registration program are the initial range and azimuth offsets between SLC-1 and SLC-2. The initial offsets can either be entered manually by identification of identical locations in SLC-1 and SLC-2 or by automatic offset estimation using program init_offset. Other parameters specify where in the image the offsets will be measured. A regular grid in range and azimuth are used for the offset locations. The grid parameters are the starting range, ending range, and number of points in range, and the starting azimuth, ending azimuth, and number of points in azimuth.

A minimum threshold for the SNR of the offset estimate is user specified. Generally, SNR values less than 5 for the CCI algorithm yields unreliable results. Similarly, SNR values below 2 for the FV algorithm are not significant. The user can specify the region of the image to generate an interferogram. The coordinates of the image section selected to be processed are relative to the geometry of SLC-1.

SEE ALSO
typedef_ISP.h, SLC_par, OFF_par, init_offset, offset_pwr, offset_SLC, offset_pwr_tracking, offset_fit, SLC_intf , WSS_intf.