A supervision learning method based on the central loss function has been proposed to enhance the recognition performance of Synthetic Aperture Radar(SAR)images in small-sample sce-narios.This method involves learning category centers for each class and penalizing the distance be-tween the deep features of samples and their respective category centers,thereby improving both inter-class resolution and intra-class dispersity.To validate the effectiveness of this approach,it is compared with common deep learning algorithms on the MSTAR image recognition dataset.The experimental re-sults show that,compared to other deep learning models,this method exhibits more superior image recognition performance in scenarios with small samples.
关键词
合成孔径雷达/小样本图像识别/中心损失函数/深度学习
Key words
synthetic aperture radar/small sample image recognition/center loss function/deep learning
维普
万方数据