Research Of Recognition On Inverse Synthetic Aperture Radar Image

Inverse synthetic aperture radar (ISAR) is a high-resolution imaging radar, with the characteristics of radar imaging at long range, with high accuracy, and under all weather conditions. It can achieve imaging on aircrafts, ships, spatial targets (such as celestial bodies, satellites, etc.) and terrain. The identification and classification of targets by radar images is extremely significant in the area of national defense and is also a hot research area nowadays. Although there are some progresses results in this field, breakthrough progress has not yet been achieved.Due to the targets' state, interference, echo signal to noise ratio, moving compensation impact and other effects, there are always some defects in the normal ISAR target image, influencing target feature extraction. In this paper, the aircraft ISAR target image was studied in an effective way by the use of curve fitting and removal of fringes and dot interference.In order to obtain high-quality features in feature extraction, this paper calculated the minimum objective coverage of the round and round in the mapping units, making ISAR target itself independent of translation and scale.This paper utilize Zernike moments to extract target image features. Zernike moments are well-defined image characterization of the shape. From the perspective of the theory, when the factorial of the demanding Zernike moments tends to be infinite, we can fully restore the original image. Besides the translation and scale invariance of the ISAR target itself, the features extracted by Zernike moments has the characteristics of the rotation invariance, which makes the identification of the target extremely favorable.In the process of target identification, we utilize K-nearest neighbor rules on target classification, the experiment on the selected target aircraft model of ISAR image recognition achieved good recognition results.