自适应训练的高分辨率SAR图像的船舶小目标检测方法  

Adaptive training for ship small target detection in high-resolution SAR image

作　　者：潘婕 PAN Jie(School of Artificial Intelligence,Ningbo Polytechnic,Ningbo,Zhejiang 315800,China)

机构地区：[1]宁波职业技术学院人工智能学院,浙江宁波315800

出　　处：《计算机应用文摘》2024年第16期164-166,169,共4页Chinese Journal of Computer Application

基　　金：2021年浙江省教育厅一般科研项目(Y202147455);2023年宁波职业技术学院校级课题项目(NZ23Z01)。

摘　　要：文章介绍了船舶目标检测的方法。针对SAR图像中船舶小目标检测因特征易消失而导致的检测困难问题,在单阶段目标检测框架的基础上,首先构建了一个基于图像金字塔残差网络的特征提取网络;其次增加了自适应训练样本选择网络,并利用难样本挖掘的思想改进了模型优化的损失函数。整体上,这些改进提升了目标检测模型对船舶小目标的检测性能。实验结果表明,所提方法能有效提升小目标的检测效果。The article introduces the methods of ship target detection.In response to the difficulty of detecting small ship targets in SAR images due to the easy disappearance of features,a feature extraction network based on an image pyramid residual network was first constructed on the basis of a single-stage object detection framework.Secondly,an adaptive training sample selection network was added,and the loss function of the model optimization was improved using the idea of difficult sample mining.Overall,these improvements have improved the detection performance of the object detection model for small ship targets.The experimental results show that the proposed method can effectively improve the detection performance of small targets.

关 键 词：小目标检测 自适应训练 SAR图像 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]