摘要
针对现有分类方法在遥感图像分类方面存在总体分类精度低的问题,引入频域Transformer,开展小样本遥感图像自动分类方法设计研究。变换小样本遥感图像频域,并实现频域对齐;对小样本遥感图像进行学习训练,利用频域Transformer对小样本遥感图像特征进行融合,实现图像超像素分割与自动分类。通过对比实验证明,新的分类方法可以实现对遥感图像的高精度分类。
In view of the low overall classification accuracy of remote sensing images in existing classification methods,frequency domain Transformer is introduced to carry out research on the design of automatic classification methods for small sample remote sensing images.Transform the frequency domain of small sample remote sensing image and realize the frequency domain alignment.Learn and train small sample remote sensing images,and use frequency domain Transformer to fuse features of small sample remote sensing images to achieve image superpixel segmentation and automatic classification.The comparison experiment proves that the new classification method can realize the high precision classification of remote sensing images.
作者
韩镇阳
陈浩
曾成
HAN Zhenyang;CHEN Hao;ZENG Cheng(Shaanxi Provincial Corps of the Chinese People's Armed Police Force,Xi'an,Shaanxi Province,710116 China)
出处
《科技资讯》
2025年第4期74-77,共4页
Science & Technology Information
关键词
频域
小样本
遥感图像
自动分类
Frequency domain
Small sample
Remote sensing image
Automatic classification
