Abridged Shape Matrix Representation for the Recognition of Aircraft Targets from 2D ISAR Imagery

摘要

Inverse Synthetic Aperture Radar (ISAR) images are used (i) for detecting the high scatter parts or areas in the target (hot-spots) and (ii) for the identification of targets. Automatic Target Recognition (ATR) systems reduce the burden of the human operators by identifying the targets in ISAR images automatically and aid the operators in executing accurate decisions with less effort. Even though target classification or recognition is the main objective these systems, they must possess the mechanisms to cope up with the noise as the clutter and noise degrade the overall performance of these systems significantly. In this paper a new classification method is proposed to identify the aircrafts in ISAR images automatically. A Homomorphic wavelet based step-by-step systematic noise removal procedure is developed, to segment out the target carefully from background noise. Rotation and scale normalization are performed on the segmented image using feature matching. Finally, the classification step depends on the abridged shape matrix generated from the segmented target to recognize the aircraft. Experiments were conducted on synlSAR aircraft image dataset and results are presented. The experimental results show that the proposed method is fast and accurate to identify aircrafts in ISAR images.