| 顾及局部与结构特征的稀疏多项式逻辑回归高光谱图像分类方法 |
| |
| 引用本文： | 沈宇臻,官云兰,杨禄,刘承承,严小芳.顾及局部与结构特征的稀疏多项式逻辑回归高光谱图像分类方法[J].测绘通报,2019,0(6):24-28. |
| |
| 作者姓名： | 沈宇臻 官云兰 杨禄 刘承承 严小芳 |
| |
| 作者单位： | 广州城市规划技术开发服务部,广东 广州,510030;东华理工大学测绘工程学院,江西 南昌,330013;桂林理工大学测绘地理信息学院,广西 桂林,541004;成都理工大学地球科学学院,四川 成都,610059;兴宁市国土资源局,广东 兴宁,514500 |
| |
| 基金项目： | 国家自然科学基金（41401437） |
| |
| 摘 要： | 稀疏多项式逻辑回归在分类中仅利用图像光谱信息,导致分类效果不太理想.本文提出了一种顾及局部与结构特征的稀疏多项式逻辑回归高光谱图像分类方法.首先利用加权均值滤波与拓展形态学多属性剖面对原始高光谱图像进行局部与结构特征提取;然后对二者进行加权平均特征级融合以获取更具唯一性的像元特征;最后由稀疏多项式逻辑回归分类器对融合结果进行分类.结果表明,本文方法能有效地提高分类精度,而且具有较强的稳健性.
|
| 关 键 词： | 高光谱影像 特征融合 加权均值滤波 EMAPs 稀疏多项式逻辑回归 |
| 收稿时间： | 2018-11-29 |
Hyperspectral image classification based on local and structural feature with sparse multinomial logistic regression |
| |
| SHEN Yuzhen,GUAN Yunlan,YANG Lu,LIU Chengcheng,YAN Xiaofang.Hyperspectral image classification based on local and structural feature with sparse multinomial logistic regression[J].Bulletin of Surveying and Mapping,2019,0(6):24-28. |
| |
| Authors: | SHEN Yuzhen GUAN Yunlan YANG Lu LIU Chengcheng YAN Xiaofang |
| |
| Institution: | 1. The Department of Guangzhou Urban Planning Technology Development Services, Guangzhou 510030, China;
2. Faculty of Geomatics, East China University of Technology, Nanchang 330013, China;
3. College of Geomatics and Geoinformation, GuiLin University of Technology, Guilin 541004, China;
4. College of Earth Science, Chengdu University of Technology, Chengdu 610059, China;
5. Xingning Municipal Bureau of Land and Resources, Xingning 514500, China |
| |
| Abstract: | Sparse multinomial logistic regression only uses spectral information in image classification.Therefore,the classification effect is poor. In this paper,a hyperspectral image classification method based on local and structural feature with sparse multinomial logistic regression is proposed. Firstly, weighted mean filter and extended multi attribute profiles are used to extract local and structural features of the original hyperspectral images. Then the weighted average feature level fusion is carried out to obtain more unique pixel feat ures.Finally,the fusion results are classified by sparse multinomial logistic regression.The results show that the proposed method has well classification effect and robustness. |
| |
| Keywords: | hyperspectral image feature fusion weighted mean filter extended multi-attribute profiles sparse multinomial logistic regression |
| 本文献已被 维普 万方数据 等数据库收录！ |
| 点击此处可从《测绘通报》浏览原始摘要信息 |
| 点击此处可从《测绘通报》下载免费的PDF全文 |
|
