Research On ISAR Imaging And Physical Feature Extraction Of Midcourse Ballistictarget

Researching on the wideband radar signal processing of midcourse target has practical significance for discriminating warheads from decoys. The dissertation focused on the research of the ISAR imaging and physical feature extraction based on wideband radar. Three main subjects have been researched in this dissertation, which are ISAR imaging, geometry feature extraction and micro-motion feature extraction.In the research of ISAR imaging of midcourse target, four issues are studied including high velocity compensation, phase auto focus, signal decomposition and cross scaling of ISAR image. Firstly, the dissertation analyzes the effect of high velocity on the HRRP and ISAR image. With the aid of chirplet presentation, the radius velocity is estimated and the phase item caused by high velocity is compensated. For phase auto focus, scattering centers are separated based on windowed rough ISAR image and expanded HRRP is constructed to get more stable Doppler center. The phase error is compensated based on expanded HRRP. For the appendix part and body part signal decomposition, single frequency radar echo is constructed. Based on the Doppler difference, the dissertation separates the echo of body and appendix by chirplet decomposition. And the ISAR image clarity of ballistic target body is improved. Lastly, the ISAR image cross scaling problem is studied. According to the ISAR image characteristic of midcourse target, a new cross scaling method based on image registration is proposed.In the geometry feature extraction, radius length, actual length, cross section area feature extraction are studied. The dissertation studies the radius length extraction based on single HRRP firstly. By the advantage of FFT algorithm and super resolution algorithm, radius length is estimated with high precision. Furthermore, the dissertation studied the actual length extraction method based on HRRP series. According to the precession effect on the HRRP, the statistical model of radius length is constructed. And an actual length estimation based on ML algorithm is raised. Lastly, the geometry feature extraction based on ISAR image is studied. The target and background is separated based on edge extraction and image segmentation. And the scattering centers are extracted based on clustering algorithm. The cross length, radius length and target cross section area are extracted.In the micro-motion feature extraction of midcourse target, the dissertation studies the precession feature extraction. According to the relation between target posture and HRRP or ISAR image, the dissertation proposes two kinds of precession feature extraction methods based on the HRRP coefficients and ISAR image registration. According to the radius length change under the precession, the dissertation proposes an algorithm based on the minimum and maximum radius length. Lastly, according to the dominant scattering centers motion, the paper proposes two new algorithms based on HRRP series and ISAR image series. All approaches and algorithms put forward in the dissertation have been tested with simulated data, compressed field data or measured data. And experiment results show the validity.The dissertation focuses on the practical problems in the midcourse ballistic target recognition, which have great significance in ballistic missile defense. The writer hopes that the researching results are helpful for constitution of the missile defense systems and improvement of the ability of missile penetration in the future.