Minimize Generating a Multi Look Image From an SLC Product

In this example, a series of BEST tools are used to convert a single-look complex (SLC) ASAR dataset into a multi-looked (PRI-like) image.

Several steps are necessary to perform multi-looking on an SLC image, as described below:

  • slant-to-ground range re-projection in SLC format
  • oversampling (2 x 2) in SLC format
  • look detection (generate amplitude image)
  • look adding by undersampling with the desired multi-look factor

Important: Note that in the ASAR online processor, the order of the steps differs slightly from those listed above. However, there is no impact on the output result.

BEST includes all the tools necessary to achieve this conversion. For more detailed information, see the individual chapters of the User Manual corresponding to each tool.

Example Data

For this example, a subset of an ASAR Image Mode Single-look Complex product (ASA_IMS_1P) covering the city of Barcelona shall be used. The original dataset is called

ASA_IMS_1PNUPA20021017_213630_000000162010_00244_03304_0008.N1

From this identifier, the absolute orbit number (3304), track number (244 - an ascending pass), acquisition date (17 October 2002) and acquisition start time (21:36:30 UTC) are evident.

The original dataset is contained on a CD-ROM supplied from the UK Processing and Archiving Centre (UK-PAC). Processed files shall be stored on the hard disk in the “working directory” C:\Best_example\SLC_to_PRI (specified in the Setup Working Directory dialogue box accessed through the Help menu of the HMI).

Set-up

The demonstration is performed using a PC running Windows™ 2000. The Visual Basic HMI (launched by double-clicking the BEST icon on the desktop) is used throughout.

Header Analysis

The first step of a BEST processing session is nearly always analysis of the SAR data product header. The HEADER ANALYSIS tool decodes the header information of a product and writes it in a file that is used by many of the other tools in subsequent processing.

Select the HEADER ANALYSIS tool from the Data Import and Quick Look menu of the HMI.

image

When an ASAR product is selected as the input, many of the other parameters are automatically set. Click [OK] to run the tool.

When the HEADER ANALYSIS tool has finished, there should be three new files in the working directory:

1. barcelona.txt (full product header information in text format)

2. barcelona.HAN (full product header information in BEST internal format, for subsequent use by other BEST tools)

3. header_analysis.ini (the “Configuration Settings” file generated by the HMI, containing all the parameter information necessary to run this BEST tool) - see the Tutorial: Quick Look Generation and Image Extraction using BEST for Windows™ for more information about these parameters and their meanings.

Full Resolution Extraction

After header analysis, it is necessary to extract some or all of the SAR dataset and write it in the BEST internal format.

Select the FULL RESOLUTION EXTRACTION tool from the Data Import and Quick Look menu of the HMI.

image

For the remainder of this example, a subset of the full dataset shall be processed. This area of interest (AOI) is a rectangle defined by its upper left and lower right corners using the (rows, columns) reference coordinates convention. The Area of Interest dialogue box is opened by clicking the “Y” radio button in the ‘AOI spec.’ field of the main tool.

image

A sub-image of 7000 rows and 2377 columns shall be processed.

There are two outputs from this operation:

1. barcelona.XTt (the defined subset of the SLC dataset, now stored in the BEST internal format) The file extension consists of “XT”, indicating that the FULL RESOLUTION EXTRACTION tool produced the file, and “t”, to denote a complex integer data format.

As the quick look image below (generated using the QUICK LOOK GENERATION tool) shows, the extracted sub-scene is a single-look scene in slant-range geometry.

2. full_resolution.ini (the “Configurations Settings” file generated by the HMI)

image

Slant to Ground Range Re-projection

Before multi-looking can be performed, the data must be resampled onto a regular grid in the range direction.

Select the SLANT TO GROUND RANGE CONVERSION tool from the Data Conversion menu of the HMI.

image

There are two outputs from this operation:

1. barcelona.SGc (the resampled SLC image in ground range geometry in the BEST internal format) The file extension consists of “SG”, indicating that the SLANT TO GROUND RANGE CONVERSION tool produced the file, and “c”, to denote a complex floating-point data format.

As the quick look image below (generated using the QUICK LOOK GENERATION tool) shows, the output image remains a single-look scene with 7000 rows. The result of the re-projection in range has been to increase the number of columns slightly to 2614.

Important: The current version of BEST has a bug in the SLANT TO GROUND RANGE CONVERSION tool. When applied to oversampled data (i.e. if the oversampling and slant to ground range conversion steps are reversed, as they are in the ASAR online processor), it inserts additional columns of zeros in the far range that almost double the width of the output. The bug will be fixed in the next release of the software.

2. slant_to_ground.ini (the “Configurations Settings” file generated by the HMI)

image

Oversampling

The SLC data should be zero-padded prior to look detection.

Select the OVERSAMPLING tool from the Resampling menu of the HMI.

image

To achieve 2 x 2 oversampling, the ‘Output Image Ratio’ should be set to “2, 2”.

There are two outputs from this operation:

1. barcelona.OVc (the oversampled SLC image in the BEST internal format) The file extension consists of “OV”, indicating that the OVERSAMPLING tool produced the file, and “c”, to denote a complex floating-point data format.

As the quick look image below (generated using the QUICK LOOK GENERATION tool) shows, the numbers of columns and rows have doubled to 5228 and 14000 respectively.

2. oversampling.ini (the “Configurations Settings” file generated by the HMI)

image

Look Detection

At this point, the amplitude information can be extracted from the complex data signal.

Select the COMPLEX TO AMPLITUDE CONVERSION tool from the Data Conversion menu of the HMI.

image

There are two outputs from this operation:

1. barcelona.CAf (the amplitude image in ground range geometry in the BEST internal format) The file extension consists of “CA”, indicating that the COMPLEX TO AMPLITUDE CONVERSION tool produced the file, and “f”, to denote a floating-point data format.

2. complex_to_amplitude.ini (the “Configurations Settings” file generated by the HMI)

Look Adding

Finally, the previous oversampling has to be reversed, and at the same time the image is averaged in azimuth to derive a multi-looked image.

Select the UNDERSAMPLING tool from the Resampling menu of the HMI.

image

To achieve a PRI-like 3-look image the ‘Output Image Ratio’ should be set to “0.166, 0.5”. This scales the output image to one-sixth the number of rows (compensating for the previous 2x oversampling in azimuth and simultaneously applying 3x multi-looking, i.e. 0.5 x 0.333 = 0.166) and half the number of columns (compensating for the previous 2x oversampling in range).

The recommended ‘Filter Type’ (as used in this example) is a user-defined 2-dimensional sinc filter with its frequency set to 2 in range and 6 in azimuth. Custom “.ker” files of 11 x 11 pixels and 22 x 22 pixels can be downloaded from this page for your own use. A future version of the software will include the filter as a built-in option.

Download filter kernels to perform 3x multi-looking on SLC data

Alternatively, use the 3 x 3 Low Pass kernel (in the “.ini” file, Filter File Name = “lop_3_3.ker”) already incorporated in the software.

There are two outputs from this operation:

1. barcelona.UNf (the 3-look amplitude image in the BEST internal format) The file extension consists of “UN”, indicating that the UNDERSAMPLING tool produced the file, and “f”, to denote a floating-point data format.

As the quick look image below (generated using the QUICK LOOK GENERATION tool) shows, the number of columns has reverted to 2614, whilst the number of rows in the ‘squared’ image is 2324.

2. undersampling.ini (the “Configurations Settings” file generated by the HMI)

image

Export

The multi-looked PRI-like image is then ready to be exported, either as a binary file (thus maintaining the precision of the floating-point data), or as a TIFF file, after first using the GAIN CONVERSION tool to reduce the dynamic range of the data.