Radar Automatic Target Recognition (ATR) and Non-Cooperative Target Recognition (NCTR)
The ability to detect and locate targets by day or night, over wide areas, regardless of weather conditions has long made radar a key sensor in many military and civil applications. However, the ability to automatically and reliably distinguish different targets represents a difficult challenge. Radar Automatic Target Recognition (ATR) and Non-Cooperative Target Recognition (NCTR) captures material presented in the NATO SET-172 lecture series to provide an overview of the state-of-the-art and continuing challenges of radar target recognition. Topics covered include the problem as applied to the ground, air and maritime domains; the impact of image quality on the overall target recognition performance; the performance of different approaches to the classifier algorithm; the improvement in performance to be gained when a target can be viewed from more than one perspective; the impact of compressive sensing; advances in change detection; and challenges and directions for future research. Radar Automatic Target Recognition (ATR) and Non-Cooperative Target Recognition (NCTR) explores both the fundamentals of classification techniques applied to data from a variety of radar modes and selected advanced techniques at the forefront of research, and is essential reading for academic, industrial and military radar researchers, students and engineers worldwide.
Inspec keywords: compressed sensing; object detection; radar target recognition; military radar
Other keywords: image quality; key sensor; NATO SET-172 lecture series; compressive sensing; radar automatic target recognition; classifier algorithm; change detection; military radar researchers; ATR; NCTR; target recognition performance; noncooperative target recognition
Subjects: Optical, image and video signal processing; Modulation and coding methods; Radar theory; Radar equipment, systems and applications; General electrical engineering topics; Signal processing and detection; Military detection and tracking systems
- Book DOI: 10.1049/PBRA033E
- Chapter DOI: 10.1049/PBRA033E
- ISBN: 9781849196857
- e-ISBN: 9781849196864
- Page count: 296
- Format: PDF
-
Front Matter
- + Show details - Hide details
-
p.
(1)
-
1 Introduction
- + Show details - Hide details
-
p.
1
–3
(3)
The book is organised in nine principal chapters, based on the lectures themselves, covering different domains and different aspects of the overall topic. Chapters 2-4 consider the problem as applied to the ground, air and maritime domains, respectively. Chapter 5 describes the impact of image quality (i.e. resolution, signal-to-noise ratio) on the overall target recognition performance. Chapter 6 considers the performance of different approaches to the classifier algorithm. Chapter 7 considers the improvement in performance to be gained when a target can be viewed from more than one perspective, as well as the ways in which natural systems, such as bats, perform target recognition. Such systems have benefitted from millions of years of optimisation through the process of evolution, and perform the target recognition process in an intelligent, adaptive manner. Chapter 8 considers the impact of compressive sensing, which is a relatively novel processing approach, showing that considerable economies may be made in sampling and processing due to the sparseness of the information in the target scene. Chapter 9 describes advances in change detection, including the very powerful techniques of coherent change detection. Finally, Chapter 10 looks at future challenges and directions for future research.
-
2 Automatic target recognition of ground targets
- + Show details - Hide details
-
p.
5
–36
(32)
An overview of the fundamentals of ground target recognition using SAR has been given. There is a tendency, when discussing ground target ATR, to consider only the most complex problems consisting of very many target classes and challenging clutter environments. However, it should be borne in mind that radar ATR is not a single problem that can or cannot be solved but that it is a continuum of problems of varying degrees of difficulty and complexity all of which provide useful military capability. This was essentially the viewpoint articulated by the NATO SET111 Task Group on ground target recognition at the conclusion of that activity. Operational radar ATR systems already exist and the overarching challenge is to push forward the solution space to achieve successful operation in more difficult and complex circumstances. To achieve this, a number of key specific challenges must be overcome, some of which are identified and discussed in Chapter 10, and to a large extent it is these challenges that current researchers are tackling. Hopefully this book will inspire those reading to contribute to the effort to solve these challenges.
-
3 Automatic recognition of air targets
- + Show details - Hide details
-
p.
37
–75
(39)
RDI is an aircraft target recognition technique, which has not been widely explored, but has potential for contributing to automatic air target recognition. The use of simultaneous range and frequency data has benefits in being able to localise aircraft propulsion systems along the range profile, which is not generally possible with other NCTR techniques. For jet aircraft it can theoretically acquire HRRP and JEM data in a single dwell with the appropriate waveform, but in practice it is not normally achievable due to difficulties in designing a radar with the appropriate characteristics. RDI can measure the signatures of jet and propeller aircraft and helicopters.
-
4 Radar ATR of maritime targets
- + Show details - Hide details
-
p.
77
–126
(50)
For commercial ship traffic the automatic identification system (AIS) is obligatory for all vessels above a certain size and thus warrants their cooperative classification and even identification. However, for smaller craft and especially for non-cooperative objects with hostile intent, the classification has to rely on classical approaches of automatic target recognition (ATR), which mostly are based on radar due to its day/night and all-weather capabilities. This leads to applications such as coastal surveillance for border control, the protection of harbour installations, ship self-defence or the suppression of drug trafficking, where the classification of ships by means of ATR schemes becomes more and more important. This is especially true in times of asymmetric (terrorist) threat and piracy. With a modern high resolution radar one has the choice of two different ways of target imaging. The first is 2D imaging, either from an airborne (SAR) or from a ground-based platform (ISAR). The latter depends on the relative motion of the target itself and therefore may be difficult in the case of non-cooperative targets. The desired axis of rotation should be vertical, which may not be the case when high sea states cause strong roll and pitch motion for smaller ships. Moreover, when the hostile ship is approaching or receding on a straight course, there is no relative rotation that lends itself to ISAR exploitation.
-
5 Effects of image quality on target recognition
- + Show details - Hide details
-
p.
127
–155
(29)
Phase gradient SAR image focusing was demonstrated to provide well-focused imagery; cross-range smearing of the imagery was significantly reduced, resulting in higher probability of correct classification as demonstrated by a 20+ target model-based classifier. HDI processing was demonstrated to improve the image quality of complex SAR imagery; the effective resolution of SAR imagery was shown to be increased as demonstrated by the improved Pcc achieved by a ten-target template-based classifier. 2D FFT image formation processing of interrupted SAR phase-history data was shown to yield SAR imagery containing significant artefacts and degraded image quality; CS-based image formation processing (BPDN) was shown to mitigate these image artefacts and produced complex SAR imagery having excellent image quality.
-
6 Comparing classifier effectiveness
- + Show details - Hide details
-
p.
157
–175
(19)
The classification and even, if possible, identification of any object or target under observation unambiguously at a distance and in such short times that a decision and an `adapted' reaction are possible was, is and will be a most important task in the military and also security domains. There is an ongoing effort in many countries all over the world, and of course also in NATO, to improve the existing solutions. One approach, the obvious one, is the idea to allow any object entering observed areas to identify itself actively on request. Such systems have been installed for many years in the military domain and are called `Q&A' systems or `identification friend or foe' (IFF) systems. For air targets, and if radar is used as the sensor system, aircraft have been equipped for many years with systems like Mark X or Mark XII in different modes which were able to answer an interrogating pulse from the observer with a predefined code revealing, in addition to other information, its identity. Systems like that are also used in the civilian air traffic control (ATC) and because they support the interrogator they are called `cooperative identification systems'. However, they show some important drawbacks.
-
7 Biologically inspired and multi-perspective target recognition
- + Show details - Hide details
-
p.
177
–211
(35)
In this chapter we aim to exploit experience of the natural world in which echolocating mammals are able to detect and classify objects with apparent ease. These observations suggest that waveform diversity and orientation strategies play an important role. It is this hypothesis that we test and show to be valid, as confirmed through real-world radar experiments. Specifically, there is additional information contained in different perspectives of a target that can help classify it, thus boosting performance. There is also a law of diminishing returns, in that the opportunity to extract additional new information reduces as the number of perspectives increases.
-
8 Radar applications of compressive sensing
- + Show details - Hide details
-
p.
213
–236
(24)
The purpose of this chapter is to give a brief overview of the principles of CS and to show how CS may be applied in a radar system to support automatic target recognition. The chapter is organised as follows. Section 8.2 gives an introduction of the basic principles of CS. Section 8.3 presents an overview of some of the main algorithms for reconstruction of sparse signals. The application of CS to target recognition based on jet engine modulation (JEM) is described in section 8.4. Section 8.5 shows how CS may be applied to high resolution imaging of targets using inverse synthetic aperture radar (ISAR). Finally, section 8.6 gives conclusions.
-
9 Advances in SAR change detection
- + Show details - Hide details
-
p.
237
–264
(28)
This chapter presents a typical SAR imaging and data collection system (DCS). Two change detection scenes were investigated. These initial noncoherent change detection (NCCD) studies focuses on a scene containing parked vehicles (the vehicle scene) and a scene containing a subtle man-made disturbance due to people who walked in a grassy area (the racetrack scene).
-
10 Future challenges
- + Show details - Hide details
-
p.
265
–271
(7)
The idea of automatic target recognition often conjures up visions of a completely general system that is able to classify all manner of different vehicle types in the most difficult of clutter environments. This then naturally leads to some scepticism that reliable radar ATR can ever be achieved. However, in reality the concept of ATR actually describes a continuum of problems of varying degrees of difficulty from very constrained scenarios to a completely general recognition system. The aim of ongoing radar ATR research is to push the boundaries in terms of difficulty and complexity level to provide greater capability in future radar systems. Some of the key future challenges, which must be faced in order to achieve this, will now be discussed.
-
Back Matter
- + Show details - Hide details
-
p.
(1)