Abstract:The state-of-the-art deep neural networks have made a great breakthrough in remote sensing image classification. However, the heavy dependence on large-scale data sets limits the application of the deep learning to synthetic aperture radar (SAR) automatic target recognition (ATR) field where the target sample set is generally small. In this work, a metalearning framework named MSAR, consisting of a metalearner and a base-learner, is proposed to solve the sample restriction problem, which can learn a good initialization as well as a proper update strategy. After training, MSAR can implement fast adaptation with a few training images on new tasks. To the best of our knowledge, this is the first study to solve a few-shot SAR target classification via metalearning. In particular, the few-task problem is defined by analyzing the effect of available training classes on the performance of metalearning models. In order to reduce the metalearning difficulties caused by the few-task problem, three transfer-learning methods are employed, which can leverage the prior knowledge from the pretraining phase. Besides, we design a hard task mining method for effective metalearning. Based on the Moving and Stationary Target Acquisition and Recognition (MSTAR) data set, a specialized data set named NIST-SAR is devised to train and evaluate the proposed method. The experiments on NIST-SAR have shown that the proposed method yields better performances with the largest absolute improvements of 1.7% and 2.3% for 1-shot and 5-shot, respectively, over the next best, which indicates that the proposed method is promising and metalearning is a feasible solution for few-shot SAR ATR.

A Sparse Aperture ISAR Imaging and Autofocusing Method Based on Meta-Learning Framework

Sparse Aperture ISAR Imaging Method Based on Meta-CV-ADMMN

AF-AMPNet: A Deep Learning Approach for Sparse Aperture ISAR Imaging and Autofocusing.

Deep Learning Approach for Sparse Aperture ISAR Imaging and Autofocusing Based on Complex-Valued ADMM-Net

Deep Compressive Imaging with Meta-Learning.

CVAE: an Efficient and Flexible Approach for Sparse Aperture ISAR Imaging

Fast Sparse Aperture ISAR Autofocusing and Imaging Via ADMM Based Sparse Bayesian Learning

ISAR Imaging of Target Exhibiting Micro-Motion With Sparse Aperture via Model-Driven Deep Network

Joint Sparse Aperture ISAR Autofocusing and Scaling Via Modified Newton Method-Based Variational Bayesian Inference

3-D SAR Autofocusing with Learned Sparsity

AF-DCDU-Net: An Approach for ISAR Autofocus Imaging via Deep Convolution and Deep Unfolding Net

Deep learning network with new weighting strategy for ISAR image enhancement

Meta Learning-Based Approach for Few-Shot Target Recognition in ISAR Images

Few-Shot SAR Target Classification Via Metalearning.

Meta-Learner-Based Stacking Network on Space Target Recognition for ISAR Images

MTRC Compensation and Sparse Aperture ISAR Imaging Based on Sparse Bayesian Learning

Iterative Optimization-Based ISAR Imaging with Sparse Aperture and Its Application in Interferometric ISAR Imaging

Computationally Efficient Sparse Aperture ISAR Autofocusing and Imaging Based on Fast ADMM.

An Integrated Network for SA-ISAR Image Processing with Adaptive Denoising and Super-Resolution Modules

Feature learning and sar imaging method based on convolution neural network

End-to-End SAR Deep Learning Imaging Method Based on Sparse Optimization