ADTS 8-8-4 Format Description

This document describes the "8-8-4" compressed data format developed by the Automatic Target Recognition Working Group (ATRWG). The 8-8-4 data format is the format used by the MIT/Lincoln Laboratory Advanced Detection Technology (ADT) Project Image Formation Processor (IFP).

8-8-4 Pixel Types

There are two types of 8-8-4 pixels (or words): HEADER & DATA. Each 8-8-4pixel is 3-bytes long.

  1. HEADER pixels contain CONTROL information. There are three header pixels at the start of each image line and one header (or trailer) pixel at the end of each image line. Header pixels ARE NOT part of the image.
    • Pixel or word validity
    • Is pixel a header pixel or data pixel
    • Parity data
    • Frame tags
    • Line numbers
    • Polarization
  2. DATA pixels contain the SAR DATA in a special compressed complex data format known as "8-8-4". The 8-8-4 DATA format pixel is composed of three bytes.
    • Byte1--> [EXPONENT]: It is 4-bit unsigned and is found in the 4 least significant bits of the byte. The 4 most significant bits are to be ignored.
    • Byte2--> [INPHASE]: Holds the REAL integer mantissa and is 8-bit, 2's complement.
    • Byte3--> [QUADRATURE]: Holds the IMAGINARY integer mantissa and is 8-bit, 2's complement.

Typical 8-8-4 Image Data Layout

imglayout.gif (3401 bytes)

8-8-4 Image Data Record Layout

imgreclayout.gif (10167 bytes)

8-8-4 Image Data Conversion

8-8-4 image data can be uncompressed into rectangular (real, image) complex format by applying the following formula to each pixel.

where -> BSCALE = 2.44140625E-04 (i.e. 1/4096 scaling factor)
where -> BZERO = 0.0v(i.e. pixel value offset )

Image Pixel Units

The resulting UNITLESS complex pixel values are the square root of SIGMA-0. SIGMA-0 (or the backscattering coefficient) is defined as:

(Radar Cross Section (RCS), Sq-meters) / (Radar Resolution Cell Ground Area, Sq-meters)

Computing SIGMA-0

To obtain SIGMA-0 from the computed pixel values, you must multiply each complex pixel pixel value by its complex conjugate value.

Computing RCS

To obtain the RCS value in square meters, first multiply each pixel value by a supplied conversion factor. Then, multiply the result by its complex conjugate value. This conversion factor (SIGMA_0_TO_RCS) is supplied with header-2 for each 8-8-4 file.