基于注意力机制的高分辨率遥感影像地表覆盖分类模型
A classification model of land cover in high-resolution remote sensing images based on attention mechanism
付俊 1周院 2张永洪1
作者信息
- 1. 自然资源部第一航测遥感院,陕西西安 710054
- 2. 中科星图空间技术有限公司,陕西西安 710199
- 折叠
摘要
为进一步推动高分辨率遥感影像地表覆盖智能化解译工作,本文构建了基于注意力机制的遥感影像地表覆盖分类模型.该模型通过注意力模块增强了地表覆盖类别的特征表征,提高了分类的准确性.同时,对陕西省关中、陕南区域的11景资源三号卫星影像进行训练,6景资源三号卫星影像进行测试.结果表明,该模型对房屋建筑、耕地、林地分类效果较好,在高分辨率遥感影像地表覆盖分类中具有可行性.
Abstract
A classification model of land cover based on attention mechanism is constructed in this paper to further promote the intelligent interpretation of land cover in high-resolution remote sensing images.The model can be used to enhance the feature representation of land cover classes by attention mechanism and improve the accuracy of classification.Eleven images from ZY-3 satellite are trained and six images from ZY-3 satellite are tested in Guanzhong and Southern regions in Shaanxi Province.The results show that the model is effective in the classification of housing and buildings,cultivated land and forests,and it is feasible in land cover classification of high-resolution remote sensing images.
关键词
深度学习/高分辨率/地表覆盖分类/注意力机制Key words
deep learning/high-resolution/land cover classification/attention mechanism引用本文复制引用
出版年
2024
NETL
NSTL
万方数据