摘要
遥感影像随着遥感技术的进步,其分辨率不断提高,所包含的信息愈加丰富,传统的分类方法无法适应复杂样本的遥感影像分类。为提升高分辨率遥感影像分类性能,该文对深度学习中的卷积神经网络进行研究,以残差结构为基础构建ResNet50网络,并在高分辨率影像数据集AID和RSSCN7上进行实验。实验表明,ResNet50网络在数据集AID上的总体准确率为90.23%、召回率为89.76%;在RSSCN7上的总体准确率为87.96%、召回率为87.96%。这表明深度学习方法在高分辨率遥感影像分类任务中能够保持较高的分类性能。
With the advancement of remote sensing technology,the resolution of remote sensing images continues to improve and the information they contain becomes increasingly abundant.Traditional classification methods cannot adapt to the classification of remote sensing images of complex samples.In order to improve the classification performance of high-resolution remote sensing images,this paper studies convolutional neural networks in deep learning,builds a ResNet50 network based on the residual structure,and conducts experiments on high-resolution image datasets AID and RSSCN7.Experiments show that the overall accuracy rate of ResNet50 network on dataset AID is 90.23%and the recall rate is 89.76%;on RSSCN7,the overall accuracy rate is 87.96%and the recall rate is 87.96%.This shows that the deep learning method can maintain high classification performance in high-resolution remote sensing image classification tasks.
出处
《科技创新与应用》
2025年第35期72-74,79,共4页
Technology Innovation and Application
基金
江西省教育厅科学技术研究项目(GJJ2403803)
赣东学院院长基金项目(12224000613)。
