| 面向对象的高光谱遥感影像稀疏表示分类 |
| |
| 引用本文： | 徐锐,林娜,吕道双.面向对象的高光谱遥感影像稀疏表示分类[J].测绘工程,2018(4):71-75,80. |
| |
| 作者姓名： | 徐锐 林娜 吕道双 |
| |
| 作者单位： | 重庆交通大学 土木工程学院,重庆,400074 |
| |
| 基金项目： | 重庆市教委科技项目资助 |
| |
| 摘 要： | 稀疏表示用于高光谱遥感影像分类多是基于像素层次来处理的。文中提出一种面向对象的高光谱遥感影像稀疏表示分类方法。首先从高光谱影像中提取4个波段组成标准的多波段影像,进行面向对象的影像分割;然后计算各对象在各波段上的光谱均值,并选取少量样本进行训练;最后利用基于Fisher字典学习的稀疏表示进行高光谱遥感影像的分类。实验结果表明,该方法可以利用较少的样本得到较好的分类效果,与基于像素层的稀疏分类相比较,分类精度与效率均有所提高,分类结果更接近真实地物,避免了零碎图斑。
|
| 关 键 词： | 高光谱遥感影像 面向对象 影像分割 Fisher字典学习 稀疏表示 hyperspectral remote sensing image classification object-oriented image segmentation Fisher dictionary learning sparse representation |
Object-oriented hyperspectral remote sensing image classification using sparse representation method |
| |
| XU Rui,LIN Na,LYU Daoshuang.Object-oriented hyperspectral remote sensing image classification using sparse representation method[J].Engineering of Surveying and Mapping,2018(4):71-75,80. |
| |
| Authors: | XU Rui LIN Na LYU Daoshuang |
| |
| Abstract: | At present,sparse representation method is used for hyperspectral remote sensing image classification based on pixel level mostly.In this paper,an object-oriented hyperspectral remote sensing images classification using sparse representation method is proposed.Firstly,four bands are extracted from hyperspectral images to form a standard multi-band image,and then the object oriented image segmentation is performed;secondly,it calculates the spectral mean of each object in each band,and selects a small amount of samples for training;finally,hyperspectral remote sensing images are classified, using Fisher dictionary learning based sparse representation.Experimental result show s that this method can use fewer samples and get better classification results.Compared with the sparse classification based on the pixel level,this method improves the classification accuracy and efficiency,the classification results are closer to the real objects,and the system patches are avoided. |
| |
| Keywords: | |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
