首页|基于Transformer的舰船高分辨距离像目标识别

基于Transformer的舰船高分辨距离像目标识别

扫码查看
基于Transformer模型出色的序列数据相关性保持以及捕获全局特征的能力,提出了一种新的基于Transformer编码器的建模方法用于舰船识别。该方法首先利用编码器部分对舰船高分辨距离像数据进行建模,然后使用位置编码来记录不同序列之间的位置信息,最后利用自注意力机制对整个舰船高分辨距离像序列进行建模,通过实验数据表明,所提出的模型识别准确率达到95。98%,具有出色的识别性能。
Target Recognition of Ship High Resolution Range Profile Based on Transformer
A new modeling method based on Transformer encoders is proposed for ship recognition based on the Transformer model's excellent serial-data correlation preserving ability and the ability to capture global features.Firstly,the encoder is used to model the ship's high resolution range profile data,and then position coding is used to record the position information between differ-ent sequences.Finally,the self-attention mechanism is used to model the whole ship's high resolution range profile sequence.The experimental data show that the recognition accuracy of the proposed model is 95.98%.It has excellent recognition performance.

Transformership recognitionhigh resolution range profileself-attention mechanism

但波、宋伟健、薛永华、卢中原

展开 >

海军航空大学 烟台 264001

烟台大学 烟台 264001

Transformer 舰船识别 高分辨距离像 自注意力机制

2024

舰船电子工程
中国船舶重工集团公司第709研究所 中国造船工程学会 电子技术学术委员会

舰船电子工程

CSTPCD
影响因子:0.243
ISSN:1627-9730
年,卷(期):2024.44(7)