Aerospace and Electronic Systems Magazine July 2017 - 40

Feature Article:

DOI. No. 10.1109/MAES.2017.150273

Automatic Target Recognition in Missing Data Cases
Deoksu Lim, Chris D. Gianelli, Jian Li, University of Florida, Gainsville, FL

INTRODUCTION
Automatic target recognition (ATR) is the process where computer
algorithms are used to detect and classify objects or regions of interest in sensor data [1]. ATR algorithms have been developed for a
broad range of sensors, including electro-optic (EO), infrared (IR),
and microwave sensors (radar). A key advantage of microwave radar is its utility for all weather and time of day compared with other
types of sensors. One limitation of many radar systems' ATR performance is the poor spatial resolution in the range and cross-range
dimensions. While the range resolution of the microwave radar
system can be straightforwardly improved by increasing the system's radio frequency bandwidth, superior cross-range (or azimuth)
resolution requires an increase in antenna size. In order to achieve
resolution similar to EO or IR systems, an enormous antenna must
be used for transmission and reception, hindering microwave radar
system's general applicability to ATR problems. However, by operating a synthetic aperture radar (SAR), the large antenna requirement can be overcome, and very fine cross-range resolution can be
obtained with a modest "real" aperture antenna. Indeed, SAR has
been used to generate high-resolution 2-D or 3-D object images at
a variety of different microwave frequencies. In particular, wideangle SAR images can contain important features of an object from
a diverse set of observation angles collected by a radar system orbiting around a target or scene. These feature-rich wide-angle SAR
images are useful for ATR due to their high resolution and comprehensive coverage of the target. Interference, jamming, or data dropouts, however, are commonplace in practical SAR environments,
and result in an incomplete data set. These missing and corrupted
data cause substantial degradation in the generated SAR imagery,
hampering their utility for subsequent ATR processing especially
when data-independent image formation algorithms are used. A
prime example of these difficulties is operating a SAR system in the
very high frequency (VHF) or ultrahigh frequency (UHF) bands,
where the spectrum is frequently crowded [2].
The purpose of this work is to quantify the performance of
wide-angle SAR imaging and ATR algorithms when the data re-

Authors' address: Spectral Analysis Lab, Dept. of Electrical
and Computer Engineering, 408 New Engineering Building,
1064 Center Dr., Gainesville, FL 32611. E-mail: (lemduck1@
ufl.edu).
Manuscript received December 10, 2015, revised July 7, 2016,
October 28, 2016, and ready for publication December 15,
2016.
Review handled by D. O'Hagan.
0885/8985/17/$26.00 © 2017 IEEE
40

cord is complete, and when data are missing at 30 and 50%. We
assume that three bands (110-170, 330-390, and 480-540 MHz)
are corrupted in 30% missing data case, and the bands (70-170,
290-390, and 440-540 MHz) are interfered in 50% missing data
case.
The publicly available GOTCHA 2008 data set is used to carry
out the performance analyses [3]. The data set contains spotlight
extracted phase history data of 33 civilian vehicles and 22 reflectors observed from 31 circular orbits with different elevation angles and a diameter measuring 5 km. The center carrier frequency
and bandwidth of the GOTCHA radar are 9.7 GHz and 600 MHz,
respectively.

ATR ALGORITHM
Image formation is a crucial step in ATR systems, especially when
data is missing or corrupted. A hybrid high-resolution SAR imaging method (see [12]) is applied to counteract the negative effects
of data loss. Following image formation, the pose, or orientation of
a given target must be corrected. A high-performance pose estimation approach using a 2-D fast Fourier transform (FFT) [9] and an
application of the projection slice theorem (PST) [10] are used. Finally, a method of extracting features from the target imagery using
the cumulative sum vector (CSV), and target classification via the
local learning based feature selection method [11] is applied. The
ATR process is displayed in Figure 1.
This article shows that recognition rate obtained from the hybrid high-resolution SAR imaging method yields better than that
from the back-projection (BP) method [8]. Each of these key steps
is described in detail in the following sections.

AUTOFOCUSING IN WIDE-ANGLE SAR
Prior to image formation, the prominent-point processing (PPP)
method [4] is used to correct a time-varying slant-range error.
This method uses a quad-trihedral object as shown in Figure 2
as a reference point, and attempts to estimate the slant-range error

Figure 1.

ATR process between hybrid high-resolution and BP imaging methods.

IEEE A&E SYSTEMS MAGAZINE

JULY 2017

Table of Contents for the Digital Edition of Aerospace and Electronic Systems Magazine July 2017