An Effective way of recognizing Multi-view Synthetic Aperture Radar Automatic Target using Wavelets based Convolutional Neural Network

Abstract

Synthetic aperture radar (SAR) is a significant microwave detection instrument for remote sensing and reconnaissance as well as automatic target recognition (ATR) in both military and civilian domains.  It is desirable for discovering optimal SAR platform flight paths as well as obtain sequence of SAR images from appropriate views which lead multi-view SAR ATR to perform accurately and efficiently. For this, Shannon M-Band-based Wavelets technique and Evolution of Cub to Predator- Convolutional Neural Network (ECP-CNN) based optimization framework to multi-view SAR ATR are proposed and implemented in this paper. Here, Shannon M-Band-based wavelets technique is used to feature extraction process. The proposed technique is used to find the optimal flight paths whereas corresponding imaging viewpoints can be acquired. At last, accurate recognition results are acquired under multi-view SAR images. Extensive experiments expressed superiority and validity of proposed optimization framework of multi-view SAR ATR.