Abstract:Without sufficient data, the quantity of information available for supervised training is constrained, as obtaining sufficient synthetic aperture radar (SAR) training data in practice is frequently challenging. Therefore, current SAR automatic target recognition (ATR) algorithms perform poorly with limited training data availability, resulting in a critical need to increase SAR ATR performance. In this study, a new method to improve SAR ATR when training data are limited is proposed. First, an embedded feature augmenter is designed to enhance the extracted virtual features located far away from the class center. Based on the relative distribution of the features, the algorithm pulls the corresponding virtual features with different strengths toward the corresponding class center. The designed augmenter increases the amount of information available for supervised training and improves the separability of the extracted features. Second, a dynamic hierarchical-feature refiner is proposed to capture the discriminative local features of the samples. Through dynamically generated kernels, the proposed refiner integrates the discriminative local features of different dimensions into the global features, further enhancing the inner-class compactness and inter-class separability of the extracted features. The proposed method not only increases the amount of information available for supervised training but also extracts the discriminative features from the samples, resulting in superior ATR performance in problems with limited SAR training data. Experimental results on the moving and stationary target acquisition and recognition (MSTAR), OpenSARShip, and FUSAR-Ship benchmark datasets demonstrate the robustness and outstanding ATR performance of the proposed method in response to limited SAR training data.

Feature separability analysis for SAR ATR using data description method

An SAR ATR Method Based on Scattering Centre Feature and Bipartite Graph Matching

Automatic Sar Target Recognition Based on Two-Dimensional Locality-Preserved Maximum Information Projection

An Efficient Sar Atr Approach

Recognition of Targets in SAR Images Based on a WVV Feature Using a Subset of Scattering Centers

SAR Imagery Feature Extraction Using 2Dpca-Based Two-Dimensional Neighborhood Virtual Points Discriminant Embedding

Sparse Coding of 2D-Slice Zernike Moments for SAR ATR

2dpca-Based Two-Dimensional Marginal Sample Discriminant Embedding For Sar Atr

SAR ATR Method with Limited Training Data via an Embedded Feature Augmenter and Dynamic Hierarchical-Feature Refiner

A SAR Target Recognition Method Based on Central Moment Features

An Effective Method For Sar Automatic Target Recognition

Exploring Reinforced Class Separability and Discriminative Representations for SAR Target Open Set Recognition

2DPCA-based Two-dimensional Maximum Interclass Distance Embedding for SAR ATR

A Recognition Method of Aerospace Time-Sensitive Targets Based on Radar High Resolution Range Profile

Multiview Feature Extraction and Discrimination Network for SAR ATR

Adaptive multiparameter spectral feature analysis for synthetic aperture radar target recognition

Kernel Marginal Sample Discriminant Embedding for SAR Automatic Target Recognition

Novel SAR Automatic Target Recognition Method Based on Contour Feature

Sar Target Recognition Via Supervised Discriminative Dictionary Learning and Sparse Representation of the Sar-Hog Feature

SAR AUTOMATIC TARGET RECOGNITION BASED ON KPCA CRITERION [J]

Sample Discriminant Analysis for SAR ATR.