| Inverse Synthetic Aperture Radar(ISAR) imaging for ship target mainly utilizes the relative motion between the ship and the radar to form a two-dimensional image. ISAR can work in all kinds of weather whitout rest; it also has both military and civil values, such as military reconnaissance, identification and detection of surface targets, technical support and guide for sea vessels. The main content of this article are as follows:1. Three kinds of different ships’ geometrical lattice models are used to describe the scattering structure of ship targets. Three kinds of ships are: Nimitz class aircraft carrier, 052 C missile destroyer, "Rong" coastal passenger cargo ship.2. The characteristics of the ship movement are analyzed and the models of three main rotation of the ship are constructed and the traditional Range-Doppler(R-D) algorithm is studied. The simulation experiments show that R-D method can acquire a relatively good image when the ship target rotates under small angle.3. Sea clutter’s physical and statistical characteristics are analyzed, and the simulation of K-distribute sea clutter has been done. The 2-D K-distribute sea clutter’s characteristics and its interference to the ISAR imaging are particulary analyzed. The simulation experiments show that the shape parameter of sea clutter will influence the quality of ISAR imaging. As the value of shape parameter increase to a certain degree, the influence will become steady. When the sea clutter’s Doppler center reaches to the center of target, the influence of sea clutter will become to maximum.4. The pretreatment of the disturbed ISAR image and the feature extraction after the pretreatment are analyzed. The procedures of the pretreatment are as the follows: elimination of sea clutter, basic morphological processing on ISAR image. After the pretreatment we will extract the features from ISAR image. There’re two kinds of basic features we want, one of them is structure features: area, perimeter, long axis, short axis, complexity, compactness, center coordinates and so on. The other one is gray features: quality, means, coefficient of variation, standard deviation, weighted filling ratio and so on.5. Support Vector Machines algorithm and Decision Tree algorithm are analyzed. Then the extracted structural features and gray features are used to the ship target indentification. The simulation experiments show that structure feature’s recognition rate is superior to gray feature’s recognition rate. The recognition rate of structure feature is influenced by changes in azimuth, while gray feature is unrelated to azimuth. With the increase of sea clutter’s power, ship target’s recognition rate decreases. |
