Automated Ship and Small Craft Classification Tools for ISAR Imagery
Navy SBIR 2006.2 - Topic N06-122 NAVAIR - Mrs. Janet McGovern - [email protected] Opens: June 14, 2006 - Closes: July 14, 2006 N06-122 TITLE: Automated Ship and Small Craft Classification Tools for ISAR Imagery TECHNOLOGY AREAS: Information Systems, Sensors, Electronics, Battlespace ACQUISITION PROGRAM: PMA-290 Multi-Mission Aircraft, ACAT I, PMA-299 Multi-Mission Helicopter OBJECTIVE: Develop innovative and robust automated ship and small craft inverse synthetic aperture radar (ISAR) imagery target classification algorithms and tool sets. DESCRIPTION: The DoD has made major investments in the development of automatic target recognition (ATR) algorithms for both synthetic aperture radar (SAR) and high-resolution radar (HRR) modes. A number of capable ATR packages have been developed including the DARPA/AFRL Moving and Stationary Target Acquisition and Recognition (MSTAR) system, the AGRI Critical Experiment (ACE) HRR ATR, the Sandia National Laboratory real-time SAR ATR, as well as others from MIT-Lincoln Laboratory and SAIC. In comparison, relatively little has been invested into ISAR ATR tools. However, supporting work has been accomplished by the Naval Research Laboratory including their polar reformatting for improved image focusing, time-frequency and micro-Doppler exploitations, image while scan mode development and ISAR classification by parts methods. A number of SAR adaptive classification engines have been investigated including principal and independent component analysis, support vector machines and neural networks. Such methods may form the basis for an ISAR classification engine however ISAR imagery is fundamentally different than SAR imagery. The operational performance of various candidate methods as a function of training data, confusion matrices, orientation variant and invariant features, articulating components and variability within a specific target type is not known. The goal of this project is explore innovative techniques that will provide a robust adaptive ISAR classification tool to assist the radar operators rapidly and accurately classifying ships and small boats in the littoral. The classification methods should provide a high probability of correct target classification for both ships and small craft including variation within a specific class. The methods should be robust in the presence of limited signature contamination in littoral operating conditions. Ultimately the classification tool must be capable of distinguishing between targets in a large number of classes. Consideration should also be given to classification tool performance with and without application various image enhancement techniques, performance against targets types that the system has not been trained on, and the value of collateral information. PHASE I: Assess and conduct a trade-off analysis of potential ISAR classification techniques. Demonstrate the feasibility and technical merit of the proposed technique and generate a detailed system architecture description. Develop an RDT&E plan addressing performance metrics, training requirements, classification attributes, target class mix, number of classes, effects of variations within a class, image aspect, image defects, image focus, image resolution and human-system interface. PHASE II: Develop prototype technology and demonstrate that it can accomplish the goal of accurately classifying ships and small craft operating in their operational environment based on their ISAR images. Whenever possible evaluate the performance using available or sponsor provided data sets. PHASE III: Working with radar system OEMs transition the technology improvements to the Fleet. PRIVATE SECTOR USE OF TECHNOLOGY: The general methods developed could be ap-plicable to a wide range of feature classification needs ranging from those of homeland security to the medical field. REFERENCES: 2. J. LI, R. WU and V.C. CHEN, Robust Autofocus Algorithm for ISAR Imaging of Moving Targets, IEEE Transactions On Aerospace And Electronic Systems Vol. 37, No. 3 July 2001, pp.1056-1069 3. R. Lipps and D. Kerr, Polar Reformatting For ISAR Imaging, IEEE 1998 National Radar Confer-ence, Dallas, Tx, 12-13 May 1998, pp. 275-280 4. M. Bottoms, R. Lipps and V.C. Chen, ISAR Techniques for Target Identification, 2002 Combat Identi-fication System Conference, vol. 1, 3-7, June, 2002. 5. K. Kawakami, H. Tanaka and K. Yamamoto, 3D Object Recognition using ISAR Image, SICE Annual Conference in Sapporo, August 4-6,2004, pp. 204-207 6. R. Wu, Z.-S. Liu and J. Li, Time-Varying Complex Spectral Estimation with Applications to ISAR Imag-ing, Conference Record of the Thirty-Second Asilomar Conference, 1-4 Nov. 1998 pp. 14 - 18 Vol.1 KEYWORDS: Inverse Synthetic Aperture Radar; Automatic Target Recognition; Ship and Small Craft Classification; Image Classification; Image While Scan; High-Resolution Radar TPOC: (301)342-2637
|