A novel classification model for SAR image target classification, named Multi-Attention Feature Fusion Network (MAFNet) was proposed. Firstly, a multi-head self-attention mechanism was applied to the original image to capture global information. Secondly, a covariance attention mechanism was introduced to further enhance the representation of channel and spatial features. Thirdly, a shallow robust feature downsampling module was incorporated to more efficiently extract effective information from the raw image. Finally, the three attention-based features were fused to obtain more representative SAR image features. This approach overcomes the limitation of traditional convolutional neural networks, which only extract features within a local receptive field. By enhancing deep features in both channel and spatial dimensions and integrating features containing global information, the model significantly improves classification accuracy and robustness. Experimental results on the MSTAR dataset under the SOC condition show that MAFNet achieves a classification accuracy of 99.96%, outperforming other existing algorithms.
NOVAKL M, OWIRKAG J, BROWERW S, et al. The automatic target-recognition system in SAIP[J]. The Lincoln Laboratory Journal, 1997, 10(2): 187-202.
[2]
IKEUCHIK, WHEELERM D, YAMAZAKIT, et al. Model-based SAR ATR system[J]. Algorithms for Synthetic Aperture Radar Imagery III, 1996, 2757: 376-387.
[3]
ZHAOQ, PrincipeJ C. Support vector machines for SAR automatic target recognition[J]. IEEE Transactions on Aerospace and Electronic Systems, 2001, 37(2): 643-654.
[4]
CHENS Z, WANGH P, XUF, et al. Target classification using the deep convolutional networks for SAR images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2016, 54(8): 4806-4817.
[5]
SHANGR H, WANGJ M, JIAOL C, et al. SAR targets classification based on deep memory convolution neural networks and transfer parameters[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2018, 11(8): 2834-2846.
[6]
DENGJ P, ZHUY Y, ZHANGS N, et al. SAR image recognition using ViT network and contrastive learning framework with unlabeled samples[J]. IEEE Geoscience and Remote Sensing Letters, 2024, 21: 4000205.
[7]
FUZ Z, ZHANGF, YINQ, et al. Small sample learning optimization for resnet based sar target recognition[C]//IGARSS 2018-2018 IEEE International Geoscience and Remote Sensing Symposium, 2018: 2330-2333.
[8]
WILMANSKIM, KREUCHERC, LAUERJ. Modern approaches in deep learning for SAR ATR [C]//Proceedings of SPIE 9843,Algorithms for Synthetic Aperture Radar Imagery XXIII, 2016: 98430N.
[9]
SUNY C, YANK J, LIW Z. CycleGAN-based SAR-optical image fusion for target recognition[J]. Remote Sensing, 2023, 15(23): 5569.
[10]
BANASK M, KREUCHERC. A multi-view CNN for SAR ATR[C]//2024 IEEE Radar Conference (RadarConf24), 2024: 1-6.
[11]
RENH H, LIUS, MIAOL, et al. Hybrid reasoning network with class-oriented hierarchical representation for few-shot SAR target recognition[J]. IEEE Sensors Journal, 2024, 24(16): 26091-26103.
[12]
ZHANGR L, DUANY Z, ZHANGJ D, et al. An adaptive multiview SAR automatic target recognition network based on image attention[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2024, 17(1): 13634-13645.
SUNJingsen, LIZongyu, YANGSen, et al.SAR image target classification algorithm based on integrated improved convolutional block attention module[J]. Journal of Naval Aviation University, 2024, 39(4): 445-452.(in Chinese)
[15]
SANDLERM, HOWARDA, ZHUM L, et al. MobileNetV2: Inverted residuals and linear bottlenecks[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018: 4510-4520.
[16]
HOWARDA, SANDLERM, CHENB, et al. Searching for MobileNetV3[C]//Proceedings of the 2019 IEEE/CVF International Conference on Computer Vision(ICCV), 2019: 1314-1324.
[17]
ZHANGX Y, ZHOUX Y, LINM X, et al. Shufflenet: An extremely efficient convolutional neural network for mobile devices[C]//IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018: 6848-6856.
[18]
TANM X, LEQ. EfficientNet: Rethinking model scaling for convolutional neural networks[DB/OL].(2019-05-28)[2025-03-24].
[19]
HEK M, ZHANGX Y, RENS Q, et al. Deep residual learning for image recognition[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016: 770-778.
[20]
HUANGG, LIUZ, MAATENL, et al. Densely connected convolutional networks[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition(CVPR), 2017: 2261-2269.
[21]
DOSOVITSKIYA, BEYERL, KOLESNIKOVA, et al. An image is worth 16×16 words: Transformers for image recognition at scale[DB/OL].(2021-06-03)[2025-03-24].
[22]
LUW, CHENS B, TANGJ, et al. A robust feature downsampling module for remote-sensing visual tasks[J]. IEEE Transactions on Geoscience and Remote Sensing, 2023, 61: 4404312.
[23]
GAOZ L, XIEJ T, WANGQ L, et al. Global second-order pooling convolutional networks[C]//2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2019: 3019-3028.