Channel Attention Module in Convolutional Neural Network and Its Application to SAR Target Recognition Under Limited Angular Diversity Condition

In the field of automatic target recognition(ATR) with synthetic aperture radar(SAR) imagery, it is usually impractical to obtain SAR images covering a full range aspect views. When database consists limited angular diversity, can lead performance degradation SAR-ATR system. To address this problem, paper proposes deep learning-based method where channel attention modules(CAMs) are inserted convolutional neural network(CNN). Motivated by idea squeeze-and-excitation(SE) network, CAM considered help improve recognition selectively emphasizing discriminative features and suppressing ones less information. After testing various types included in ResNet18-type base SE its modified forms applied using MSTAR dataset different reduction ratios order validate improvement under diversity condition. Keywords: Synthetic Aperture Radar, Automatic Target Recognition, Convolutional Neural Network, Channel Attention Module 합성 개구면 ë ˆì´ë”, 자동 í‘œì  식별, 합성곱 ì‹ ê²½ë§, 채널 특징 집중 모듈